PRE2020 4 Group2
|Jasmijn de Joode||1358073|
|Dirk de Leeuw||1358081|
|Job van Heumen||1380036|
Natural disasters, war, terrorism or other causes, can cause structures to collapse in urban areas. When this happens, Urban Search and Rescue (USAR) will be deployed to search for and rescue the victims trapped by these structural collapses. There are certain steps to be taken when a building collapses: one of them is to analyze the building for safety hazards. A very dangerous hazard is fires and explosions, which can be caused by gas leaks.  It is important that this analysis of the building is done as fast as possible, since the chances of survival are highest when people are rescued within 72 hours . After this time frame, the number of survivors found drops drastically. Robots might be able to help meet this 72 hour deadline, by going into the building before it is safe enough for the search and rescue workers to go inside of the rubble and analyze the building for possible gas leaks.
There are different kinds of Urban Search and Rescue robots, but we will focus on one kind in particular: RoboBees. RoboBees using a particle swarm optimization algorithm may be able to help to find gas leaks in collapsed structures and thus prevent explosions and help meet the 72 hour deadline without diminishing the safety of the search and rescue workers. To be able to do this, the RoboBees have to communicate with each other and with the USAR team. How can we design the communication aspect of the Robobees in such a way that the communication between both the Robobees themselves and the Robobees and the USAR team is most efficient for the discovery of possible gas leaks in collapsed structures?
For this project, we will do research on the communication needed between RoboBees and the USAR team to be able to use a particle swarm optimization algorithm for detecting gas leaks in collapsed buildings. This consists of determining what needs to be communicated between the Robobees themselves and between the Robobees and the USAR team and how this can be communicated. We will also shortly discuss a state-of-the art of rescue robots in general; the particle swarm optimization algorithm itself; the societal relevance of the Robobees for detecting gas leaks; and lastly limitations and possible future improvements of the communication aspect when using RoboBees for detecting gas leaks.
The target group we're focusing on can be defined as the rescuers that are operating within these collapsed buildings. As mentioned in the problem statement, the USAR department is dealing with this specific problem of gas leaks and their consequences. Therefore they will make use of the RoboBees to detect any gas leaks. The use of Robobees can be extended to other departments when needed in the future. Besides the detection and contribution to the 72-hour deadline, will the rescuers benefit from the fact that the RoboBees will enter the building in their place. Buildings need to be safe from any hazards before any human being can enter them, by sending robots the danger will decrease for the USAR department itself. Interview with the user: File:Filename
Society will benefit from these robots since the time needed to inspect the building will be reduced. This will influence the time the rescuers will start to look for victims and rescue them. The chances of surviving are negatively correlated with the duration of being stuck in the building for example. This will affect the directly related victims present in the building and the neighborhood will benefit from this efficiency. The earlier the location of a gas leak is detected, the more they can prevent and avoid any consequences.
The enterprises involved in this specific problem will manufacture the materials and the development of the RoboBees. Thereafter they will be bought by the government since the USAR or fire department will be using the RoboBees to detect gas leaks within collapsed buildings.
|1||First lecture and think about possible subjects|
|2||Do research about chosen subject|
|3||Do research about specific rescue robots and specify subject.|
|Research what needs to be done in case of building collapse. Look into what needs to be done in order to be able to enter a building safely.|
|4||What does the robot need for specific techniques? Look into what techniques already exist and what is still needed.|
|5||Research communication between RoboBees and look at limitations/problems|
|6||Design prototype. Discuss the prototype and what possible future improvements are|
|7||Extra time to finish the prototype or solve other unforeseen problems. Make the presentation|
|9||Clean up wiki|
State of the art
In this section we look at examples of the newest/best-developed robots that exist at this time. The different robots have been divided into ground robots and aerial robots. The ground robots are then again split up into legged robots and tracked and wheeled robots. There are also robot competitions. What is the value of this?
We will discuss the state of the art of Search and Rescue (SAR) robots. There are two types of SAR robots, namely Urban Search and Rescue and Maritime Search and Rescue. We will only focus on Urban Search and Rescue (USAR). USAR may be needed for multiple kinds of emergencies, for example earthquakes, storms and tornadoes, floods and technological accidents. Research and projects are concerned with localizing, extracting and medical stabilization of trapped victims.
Since disaster areas are often dangerous for humans, it is convenient to make use of robots to investigate the area and help victims. Robots are also capable to carry out tasks which are very hard or even impossible for human rescue teams, for example finding victims with the help of thermographic cameras.
Current applications of USAR robots do not include autonomous robots. This is because current technology is not advanced enough to develop fully autonomous robots which are capable to cope with these complex, unpredictable and unstructured environments. Maybe in the far future there will be autonomous robots which can execute USAR missions without any help of humans. However, this is not realistic on short term. This does not mean that robots are not helpful in current USAR projects and missions, but we need to find a balance in the human-robot interaction. There are different kinds of USAR robots, we will discuss ground and aerial robots.
One of the primary challenges for ground robots is the movement in the environment. This is a hard task, because in contrast with for example traffic environments, disaster areas are often unstructured, unpredictable and unknown. It usually also contains many obstacles. To avoid these obstacles, USAR teams can make use of legged ground robots.
Legged robots are robots with jointed limbs, that often imitate legged animals. They can assist a rescue team with carrying heavy payloads, are able to perform long-duration missions, and are able to interact with the environment and to move in complex terrains. For search and rescue, they should be able to apply different gaits and maneuvers depending on the terrain and obstacles..
ANYmal is developed to be deployed on the field and work in harsh conditions. It has incorporated laser sensors and cameras, so the robot can perceive its environment, accurately localize and autonomously plan its navigation path and get there by carefully selecting footholds while walking. It only weighs about 30 kg, so the robot can be easily transported by a single operator. Researchers are still further improving the locomotion skills of ANYmal to make it capable of dealing with situations that might be encountered in a search and rescue scenario and they will integrate an arm on ANYmal: this will result in the robot having the potential to manipulate objects, to move through closed doors, or simply to use the arm as an additional point of contact.
Krock-2 is, like ANYmal, a quadruped rescue robot with sensors and subsystems to make it suitable for disaster response missions. The robot has force sensors with which the robot can feel its immediate environment, and thus improve its locomotion capabilities. It can surpass obstacles twice as high as the robot itself but also move under narrow passages of the same height. The components of the robot can be replaced easily on the field. Next to that, the robot is mechanically symmetrical from top to bottom, front to back and left to right, which results in the robot being able to operate in any condition even after a fall.
Atlas, unlike the other legged robots mentioned here, is more human than animal like and walks on two legs instead of four. Atlas is intended to aid emergency services in search and rescue operations: this includes performing tasks such as shutting off valves, opening doors and operating powered equipment in environments where humans could not survive. It is able to walk over a wide range of terrain, like snow, and can do backflips and cartwheels. It uses sensors to remain balanced, avoid obstacles, assess the terrain, help with navigation, and manipulate (moving) objects. In the 2015 DARPA competition of robotics, Atlas was able to complete all eight tasks as follows: drive a utility vehicle at the site; travel dismounted across rubble; remove debris blocking an entryway; open a door and enter a building; climb an industrial ladder and traverse an industrial walkway; use a tool to break through a concrete panel; locate and close a valve near a leaking pipe; connect a fire hose to a standpipe and turn on a valve.
Another quadruped robot is BigDog, which is not created particularly for rescue, but for military use. It was created in the hope it would be able to serve as a robotic pack mule to accompany soldiers in rough terrains, instead of conventional vehicles: because instead of wheels or treads, BigDog has four legs, allowing it to move across surfaces that are too rough for wheels. BigDog uses a variety of sensors, including joint position and ground contact. BigDog also features a laser gyroscope and a stereo vision system. It can travel on several kinds of terrain, like ice, mud, forest and it is able to recover balance after skidding in slope or when kicked by someone. By the hand of this robot, Boston Dynamics has developed more quadruped robots like WildCat (which was the fastest untethered quadruped robot in the world in 2013), LittleDog, and Spot.
Another quadruple-legged robot is MIT Cheetah: the total power utilized by the robot is very much similar to running animals. The robot has good locomotion skills, it can for instance run on a treadmill and on grassy and uneven terrain in a controlled manner and jump over hurdles.
Why (four-)legged robots?
Legged robots have more potential than wheeled or and tracked robots because they can work in cluttered terrain, complex and hazardous environments, since they are more like humans and animals. The quadruped robots are the best choice among all legged robots related to mobility and stability of locomotion, because four legs are easily controlled, designed, and maintained compared to two or six legs. However, legged robots in general are hard to control. These robots need to have complicated structures with multiple legs and actuators for stability and multiple sensors to perceive their environment. This results in that the legged robots are often more expensive than wheeled or tracked robots. This is one of the reasons that legged robots are currently not used as much as wheeled or tracked robots in real-life situations. Another technical problem is that legged robots in a disaster environment need to be very adaptable, not just in their intended design, but they also need to be able to adapt to possible damage, for example to the loss of one of its legs. This is important, because legged robots are usually more fragile than wheeled or tracked robots. Learning algorithms can be used to solve this problem.
Tracked and Wheeled robots
One of the largest problems with Rescue robots is mobility. Many robots already in use outside of USAR can only move on certain terrain types mostly on flat surfaces. But in USAR this is not always an option. So for wheeled and tracked robots we need options that allow us to traverse difficult terrain. A few options of The wheeled and tracked robots are:
The triSTAR locomotion unit - This robot consists of base with 4 wheels. But to overcome the mobility problem it has special wheels. These wheels are actually three wheels that turn individually and if an obstacle is met these three can turn around each other to climb the rubble. This topic has been researched already and i do not think that many more advances can be made on this robot.
CMU modular snake - A second option would be the CMU modular snake. These Robots consists of smaller parts that can move independently from each other. THis allows this robot to move through difficult terrain. This robot can still be developed further and may be interesting to look at. We also have more types of modular snake robots like the kulko, PIKo and Omnithread robots.
Stanfords snake robot- This robot works by extending from the tip. This allows the robot to travel through loose sand and small holes and even spikes. Since this robot is still in development many advances can be made.
Unmanned Aerial Vehicles (UAV’s) have many benefits over ground vehicles. They can be used to get a good view on the disaster area, but they can also be used to get to small spaces where UGV’s cannot reach. However UAV’s also have some disadvantages. Because of their size and power constraints, UAV’s are not able to carry heavy equipment or medical supplies with them. Furthermore, they are usually quite fragile. This means that UAV’s need to be very careful to not collide into walls or other obstacles.
When a building collapse, the first hours are the most crucial. The search and rescue is extremely time-consuming and difficult at the same time. The main reason what makes it this hard, is because it is simply too dangerous to just enter. In these cases, a tunnel needs to be constructed or the building needs to be entered from the top . The presence of living victims demands the available resources on those structures. Unmanned aerial vehicles can provide a solution to this time-consuming problem . UAV’s are capable of entering small spaces and search through structures that are too dangerous for the rescuers to enter. These small vehicles can pinpoint the exact location of the victims within the collapsed building, and potentially detect the assessment of the victims’ condition. When using UAV’s in collapsed building search mission, a highly specialized equipment is needed. The following list shows the must haves for such a mission and some considerations. Note that most of the features add some weight or require power, only limited supply available!, can affect the endurance of the UAV.
Four type of drones: 1. Multi-rotor: best option for usar! 2. Fixed wing 3. Single roto 4. Hybrid vtol
ICARUS focuses on developing integrated tools for search and rescue, utilizing teams of air, ground, and marine vehicles. One of the projects of ICARUS is an unmanned ground vehicle (UGV) which consists of two UGV’s: One large UGV and one small UGV. The larger UGV is used as a Mobile sensor platform. The large UGV collects large amounts of data that is necessary to navigate through the environment. Platform for powerful manipulator. The large UGV will be able to remove small obstacles from its path or to free a victim. Transport platform for small UGV. The large UGV will be used as a platform to carry several small UGV’s. The small UGV is used to enter collapsed buildings without damaging these buildings. These small UGV’s can be used to find and help trapped or injured victims. Since this vehicle is small it cannot be equipped with sophisticated sensors nor with a powerful computation unit. For this reason, the level of autonomy of this vehicle will be quite low.
After the earthquake in Amatrice, Italy in 2016, a team of the TRADR project made use of two UGV’s and three unmanned aerial vehicles (UAV’s) (see the picture on the right) to provide a 3D model of two partially collapsed churches. The mission was a success and the UGV’s and UAV’s were able to collect enough data for a high quality 3D model of the two churches.
Natural disaster scenarios are one of the reasons why ground robots are being developed. One of the main challenges is to be able to navigate on complex terrain. Another reason that pushes the technological development of tracked and wheeled robots are open robotic challenges. Some examples are: The ARGOS challenge (ARGOS, 2017). This challenge was won by team argonauts with a tracked robot from the company TAUROB (TAUROB, 2017) (see Figure 3). The DARPA Robotic Challenge. After the nuclear disaster at Fukushima in Japan 2011 it became clear that humans are very vulnerable to natural and man-made disasters. Existing rescue robots were at that time unable to prevent or reduce the damage. This was the reason that the Challenge was created in 2012 by the Defense Advanced Research Projects Agency (DARPA). The primary goal of the challenge was to stimulate the development of human-supervised ground robots which should be able to execute complex tasks in dangerous environments.
What to do in case of a structural collapse?
As the title pretty much already tells us, this section is about what to do in the case of a collapsed building. We already saw that in a rescue mission, the first 72 hours are the most important for rescuing survivors. In this section we will see what needs to be done to act as safely and efficiently as possible.
There are multiple sources that give a 5 step plan in case of a collapsed building. Although they differ slightly, they all boil down to the same 5 steps. We will use the steps set by the OCHA
- The first step is initializing the search and rescue teams as quickly as possible.
- The second step is analysis of the building. What are the dangers for rescue workers when entering the building?
- The third step is finding the survivors in the building.
- The fourth step is getting people out of the rubble.
- And the fifth and last step deciding when to close the operation.
Step 1 - initializing search and rescue teams
On the site of USAR, it says that they are available in the Netherlands within 4 hours and internationally they can be on-site within 24 hours. As we saw in the problem statement, it is very important for search and rescue teams to be on-site and working as quickly as possible, since the first 72 hours are the most important in recovering survivors.
Step 2 - analyse the building
The book Protecting Emergency Responders, Volume 4: Personal Protective Equipment Guidelines for Structural Collapse Events gives a great overview of the different dangers there are in a search and rescue mission. So, we will use this book to give a short summary of the different kinds of hazards. They make the differentiation between physical, chemical and biological hazards. These three hazards are in turn divided into sub hazards. So, we will look at these separately, since they all need different approaches to dealing with them.
Falling objects and collapsing structures
Electric shock- After a collapse, electric cables can be downed and severed. Rescue workers and people who were in the building at the moment of the collapse can get hurt from this in several ways. Rescue workers can get hurt from direct contact with an electrical source, but electricity can also reach a rescue worker through the air. Possibly clothing can catch fire because of heat generated by an electric source. And of course, if there is flooding, electricity could also travel through the water. Basically, there are many ways in which severed wires can hurt rescue workers. So, it is very important to make sure that electric lines are not energized.
Fires and explosions - There might be fuel stored on-site or be gas leaks that provide fuel for fires or explosions. an explosion could have been the cause of the collapse in the first place, or maybe pipes have burst because of the collapse. In either case, it is important to check. Especially, since electric cables might be severed, they can easily spark a fire. Also, the collapse might be the result of a terrorist attack. In this case, there might for example still be bombs that could go off.
Hearing loss because of excessive noise - The site of a collapsed structure is very noisy. For example, in order to free people from the rubble, excavating equipment is needed. These tools for drilling and digging make a lot of noise.
Asphyxiation hazard - There are a number of reasons why there might not be enough oxygen in a certain space. There can be oxygen consumption or oxygen displacement. Oxygen consumption can happen when there is combustion in a poorly ventilated space or if there are a lot of people in a small space. Oxygen displacement can occur when large amounts of gasses are released into a (small) space.
There are numerous chemicals that can be released due to the collapse of a building. The chemicals can come out of the building materials that are pulverized or chemical storage tanks or containers can be damaged because of the collapse. If there is incomplete combustion or fires, this can increase the amount of chemicals in the air.
The last hazard to discuss is biological hazards. Possible ways for pathogens to be released are from damaged sewer systems or they can be bloodborne from infected patients.
Step 3 - find survivors
This step has been most researched in combination with robot assistance. On multiple occasions, robots have helped find people caught in the rubble.
Here we will mainly look at ways where no robots are used.
Rescue dogs - The use of rescue dogs is very important. The exceptional nose of dogs can help the rescue workers locate victims. One of the downsides of using dogs is the communication between the dog and its handler. This lack of communication might lead to incomplete or inaccurate information, which in turn could mean missing someone. . Another problem is that dogs might get to a location where rescue workers are not (yet) able to get. This is why research has already been done as to how robots might assist canine search.
Knowledge of locals - Locals might know where in the building the chances are the biggest that people were there at the time of the collapse.
Weak buidings - Weak buildings can of course collapse very easily, but the material is often also very light. Which means that the chance of people surviving even though they are covered under the building, is higher than when a building is made of very heavy material.
Strong buildings - Certain parts of buildings are stronger than other parts. Under/in parts that are very strong, the chances are largest that there are open spaces where people are caught in. So, if people were inside these spaces at the time of the collapse, the possibility of survival is quite high.
Video cameras - Using cameras with heat vision, the hot spots on the video might be identified as people.
Listening - This is a powerful tool as people who cannot move but are still conscious might be shouting to get people to hear and locate them.
Shifting rubble - When lifting debris, limbs of survivors can become visible.
Step 4 - rescue survivors
Once the hazards of going into the building have been identified and it is known where survivors are located, the next step is to rescue the survivors. The only way to do this is by creating pathways through which the victims can be reached and help them get out.
Step 5 - close the operation
At some point, the chances of finding survivors has become so low, that the operation can be closed. But deciding when this is, is always a difficult decision to make.
For the remainder of this project we will look at how robots can help in the second step - analysis of the building. For the first and last two steps, people are needed. At least for the foreseeable future, our technology is not advanced enough for robots to do this on their own or be of significant help in these steps. For step 4 - finding survivors - a lot of options already exist, such as infra red video search and robots that can assist dogs. So, we think that there is room for improvement for robots that can help survey the building to make sure rescue workers can enter the collapsed building safely.
After we had done research about USAR robots in general, we found out that we needed a more specific and clear problem statement. We already discussed some possible problems statements, for example:
- How can robots help or free trapped victims?
- How can robots stabilize buildings after a disaster?
- How to find out whether a building is safe enough to enter?
- How to find or even stop possible gas leaks?
After some discussion, we decided to focus on the problem of finding gas leaks. We think that this is a specific and clear problem. After this, we tried to find out which robots would be best suited for this task. It is important that the robot is able to deal with complex terrain and can enter small spaces. We considered the following options:
- Wheeled or tracked robots. Wheeled and tracked robots are usually not very good at dealing with complex terrain and obstacles compared to legged robots and aerial robots. They are often also quite large. Therefore, they are probably not the best option to deal with this problem.
- Legged robots. One advantage of legged robots is that they are well suited for passing through harsh terrain. This is very useful because after a disaster the terrain can be very uneven and can contain many obstacles. Legged robots also have some disadvantages. They are, just as wheeled and tracked robots, usually quite large, so they are unable to enter small spaces. It is also a hard technological challenge to develop a well-functioning legged robot. So currently legged robots may not be the best option for this problem.
- Aerial robots. Aerial robots are compared to ground robots easier to navigate through complex terrain. They are also well suited for entering narrow spaces. A disadvantage of aerial robots is that they cannot lift heavy equipment. This also means that they cannot carry a large battery and therefore they have to recharge faster than ground robots. However, we think that the advantages, in this case, outweigh the disadvantages and therefore we think that small aerial robots are the best option for dealing with this problem.
So in the end we choose to do some research on aerial robots and we found the so-called RoboBees.
What are RoboBees?
RoboBees are flying microrobots. They are developed by researchers at the Wyss institute (a research institute at Harvard University) and researchers at the Harvard School of Engineering and Applied Sciences (SEAS). The Wyss institute develops new engineering innovations which are inspired by nature. They use the term biologically inspired engineering for this. One of the innovations of the institute are the RoboBees. The RoboBees are inspired by insects, in particular bees and flies. The researchers try to understand the biology of flies, because flies are very good at maneuvering through the air even though they have tiny brains. Besides their use in search and rescue missions, RoboBees can also be used for crop pollination, surveillance, as well as high-resolution weather, climate, and environmental monitoring. 
The RoboBee development can be divided into three main components :
- The body. Body development consists of constructing robotic insects able to fly on their own.
- The brain. Brain development is concerned with “seeing” the environment with the help of sensors and reacting to it.
- The Colony. The Colony’s focus is about using the RoboBees to work together as a swarm.
The body frame of the RoboBees is made out of carbon fiber, this is a very strong and lightweight material. The joints are made out of plastic. To construct these RoboBees, researchers at the Wyss Institute have developed innovative manufacturing methods, so-called Pop-Up microelectromechanical (MEMs) technologies. This is a technique inspired by origami. The wings of the robot flap at 120 times per second using piezoelectric actuators. These are strips of ceramic which can expand and contract with the use of an electric field . Researcher are also developing a version of the RoboBee which is made out of soft material, this makes it more flexible and less fragile .
The RoboBees are already able to do a controlled flight . This means that they are able to hover and steer with the help of a power cord. Currently, researchers are working on the development of RoboBees which are able to fly without the help of an external energy source. This is a hard challenge, because at small scales flight is quite inefficient – it costs relatively a lot of power. However, researchers already developed a RoboBee which can fly without power cord inside the lab . This version of the RoboBee had extremely lightweight power circuits, and high efficiency solar cells on board. The RoboBee needed the solar power of about three Earth suns to fly. The researchers simulated that level of sunlight in the lab with halogen lights. So, currently RoboBees are not far enough developed that they can do a solo flight outdoors without external power source, let alone with equipment like GPS and gas detection on board. The researchers will continue to develop the RoboBee until the RoboBee is able to do a solo flight outside. Since the RoboBees are very small - which makes flying quite inefficient - Harvard roboticists have to come up with solutions to save energy. They developed a version of the RoboBee which was able to stick to surfaces . This was inspired by animals like bats, birds or butterflies. The team made use of electrostatic adhesion, this is the same principle that causes a negatively charged balloon to stick to a positively charged wall. This method is about 1000 times more energy efficient compared to hovering.
Once the microrobot is able to fly without an external power source, the next step is onboard control. This means that the robot is able to “see” and react to the environment with the help of sensors. This is currently not possible, because the RoboBees cannot carry heavy sensors with them on the flight. Besides this problem, developing a robot that is able to fly on its own is hard challenge, this is the case because small changes in airflow can have a big effect on its movement. Therefore the control system of the microrobot must be able to quickly react to these external forces in order to keep the robot stable. This means that the control system has to be lightweight, small and computationally efficient. Currently, control is still wired in from a separate computer, though a research team is working on a computationally efficient brain that can be placed on the robot’s body. .
The Wyss institute is already working on programmable robot swarms. However, they are not able yet to implement it in the RoboBees. This is because the RoboBees are currently not far enough advanced - they cannot carry enough equipment with them to operate as a swarm. However, the research team at the Wyss institute developed a simple low-cost robot called “Kilobot” which is an easy to use robotic system . With these robots the researchers are able to test collective behavior and advance the development of robot swarms. For example, the researchers already developed an algorithm which allows the Kilobot robot swarm to form into any desired shape.
Requirements for RoboBees
For the swarm of robobees to work and find gas leaks some requirements need to be met. First and most simply: The swarm of robobees needs to be a swarm. Or in other words you need a lot of robots to make the algorithm efficient. The most popular number of population sizes are 40 to 50 particles. But at least 20 and at most 100 particles are needed for an optimal performance . A second requirement for the swarm of robobees would be that these robots can communicate between each other. This could either be done via Bluetooth, wireless LAN, radio waves, or similar media transmissions or communication via the environment (stigmergy) . The form of communication the robot makes use of is very important since some are larger and most importantly heavier than others. Which brings us to another requirement the robot needs to fly to be able to traverse the difficult terrain. Which means that the robots should be light enough. So the tradeoff between weight and fly time also becomes important. Although a lot of research already exists on this topic. Another important part of the robot would be the ability to run the algorithm. This could either be done by the robots itself if they are strong enough or by a central robot, a “queen” if you will. Further more the robots should be able to traverse difficult to access areas. And should thus be small and flying. Otherwise different problems arise. Also the robots should be able to detect the percentage of oxygen in the air. this would make sure that the robots are able to detect when a gas is present does not matter what type of gas.
A swarm of UAVs is a set of aerial robots that work together to complete specific tasks. The flight can either be controlled manually or autonomously by using processors on the drones. A swarm formed by multiple UAVs has many advantages compared to a single-UAV system. For detecting gas leaks in specific, the most important advantages are:
- Scalability: using multiple UAVs increases coverage.
- Speed of mission: missions can be completed faster when multi-UAV systems are used. This counts especially for search tasks, since swarms can process tasks in parallel.
- Cost: research shows missions can be completed at lower costs when swarms are used instead of single-UAV systemsCite error: Invalid
<ref>tag; invalid names, e.g. too many.
- Communication needs: Swarms only have to use one specific UAV that communicates with the center and forwards messages to other UAVs, whereas a single UAV needs to maintain contact with the ground at all timesCite error: Invalid
<ref>tag; invalid names, e.g. too many.
The idea behind swarm technology
In order to find a gas leak, we want a swarm of robots to move towards the gas leak. Logically, the closer a gas leak is, the higher the concentration of gas in the air is. So, if we let a swarm of robots measure the concentration of gas in the air and let them move towards the location with the highest concentration, we will find the leak. Therefore, we can view this as an optimization problem. There are many ways of finding a minimum or maximum. However, not all will work for this situation. First, we know nothing about the environment. Also, because air will flow through the building, the gas concentration on a specific position might not be the same every time. This means that there is some uncertainty about the value of the gas concentration that a robot might measure. Metaheuristic algorithms can help us solve this problem of uncertainty. Another reason why the use of metaheuristic algorithms is desirable is that metaheuristic algorithms have low algorithm complexity (https://www.sciencedirect.com/science/article/pii/B9780124051638000016). This means that calculations involving the algorithm can be done rather quickly.
There are many different metaheuristic algorithms. However, of many of them, it is not well understood why they work exactly (https://www.sciencedirect.com/science/article/pii/B9780124051638000016). There are a few that are better understood, such as genetic algorithms and particle swarm optimization (PSO). We see that PSO is mostly implemented in robot swarms that have been developed so far. This is because genetic algorithms use step sizes for locations. When implementing an algorithm in robots, we want to be able to be in an entire space, and not have to jump from position to position. There are many more more metaheuristic algorithms, but the ones we found are mostly about finding the shortest path to a certain goal. For us, this is not interesting, since we are not interested in the path towards the goal, but we want to find the location of the goal.
Particle Swarm Optimization
PSO uses multiple entities that communicate and cooperate with each other in order to find a maximum. Each entity keeps track of four different :
- The location where the entity currently is (a vector called x_i)
- The location where the entity has found the highest value so far (a vector called called p_i)
- The value of p_i (a number called pbest_i)
- The velocity of the entity (a vector called v_i)
Here, the i identifies which entity the vector belongs to. The PSO algorithm works in iterations. Each iteration, the value of where the entity currently is is measured and evaluated. If the value of the current location is higher than pbest_i, then p_i is updated to represent the current location and pbest_i takes the value at p_i. If the value of the current location is lower than pbest_i, then p_i and pbest_i remain the same. There is a fifth value that is kept track of. This value is called gbest. gbest represents the location with that corresponds to the highest of the pbest_i values. At the end of each iteration, the velocity of the entity is updated. The next value of v_i, depends on its current value, the personal best location of the entity (p_i), the overall best location of all the entities and it is partly random.
The way PSO can be used in gas detection is by taking multiple robots and letting these robots be the entities in the PSO model. Every robot carries a gas detection unit and the concentrations that this unit measured can be used as the values for the PSO model.
Requirements for a working swarm
Now, we look into what is needed for a swarm to successfully implement PSO. There are a few requirements that are key for a working swarm (https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=4223155&casa_token=vTClbprFmjAAAAAA:lXzejZyCbUoXExgKI91RR_SL18LGl3i8SKh3KBwm06f2uzHrynwmOYUZoykhnsRM0L9_dz4D). Every robot in the swarm must be able to
- measure the surrounding environment, such as measuring the gas concentration in the air and moving around obstacles.
- perform calculations, such as where it will go next and comparing different levels of gas concentrations in the air.
- determine accurate position information of where in the space it is exactly.
- move to a new location.
- (wirelessly) share information with each other and the person using the swarm (in our case rescue workers).
Experiments using robot swarms
There have been experiments and attempts to use PSO in robot swarms. In these experiments, the robots were often successful in finding the desired location.
Particle Swarm Optimization - In this experiment (https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=4223155&casa_token=vTClbprFmjAAAAAA:lXzejZyCbUoXExgKI91RR_SL18LGl3i8SKh3KBwm06f2uzHrynwmOYUZoykhnsRM0L9_dz4D ), a robot swarm is used to find the location of a light source in a room of approximately 3m x 3m. The light source was placed near the ceiling. Using this set-up, 6 different experiments were done with the same technology. 2 variables are varied: the number of robots that are part of the swarm (1, 2 or 3) and whether there are obstacles in the room or not. For this experiment, they used wheeled robots with a sensor that could measure light intensity. The goal was for the robots to find the spot on the ground that was exactly under the light source that was placed near the ceiling.
In order to solve this, they implemented PSO to help the robots find the light source. The light intensity that the robots measured was used as the p values in the PSO algorithm. To minimize using computing power, the robots only communicated with each other when they found a new general best value (the gbest value in the PSO algorithm). They also limited the mobility of the bots: they could not make sharp turns or go backwards. This is different from the traditional PSO algorithm, where every entity can move in total freedom.
Each of the 6 modifications of the experiment was done 10 times under the exact same condition. The lamp and obstacles would be at the exact same location and the robots would also start at the same location. Without obstacles, if two or three robots were used, the experiment ended successfully every time. The robot on its own had more trouble and only found the light source eight out of ten times. When objects were added, the robot on its own did even worse. But also, if two or three robots were used, they didn't have a perfect score anymore. After taking a look at why it didn't work, the researchers found that the location and illumination measurements had some imperfections. After correcting this, they did the experiments again. Now, if two or three robots were used, the robots found the light source correctly 100 per cent of the time, with and without obstacles. When using 1 bot, the results did not improve.
If we look at the qualitative results, we see that the path the robots take is quite inefficient. This is caused by the fact that the robots cannot make sharp turns. So, the robot circles around light source before eventually finding it.
Particle Swarm Optimization and Flocking algorithms - The second experiment (https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8023514&casa_token=qvQHJGsM36AAAAAA:J11XBCAQVEv6ENSbre-DziS15bSSZ5etfrS-dHmzRpuH-feImBpRix3GKWxS5T98M1ZVwrGk&tag=1) that we found used a bigger experiment space. The setup of this experiment was an area of 100m x 100m. They used flying robots/UAV's to find the gas leak. The UAV's would start separated from each other and also on the other side of the area than the gas leak was located. After the UAV's started flying, they would settle at 2 meters above ground.
In order to locate the gas leak, the implemented a combination of two algorithms: PSO and Flocking. They started by defining the flocking rules:
- Seperation - this rule is needed in order to avoid collisions,
- Allignment - individuals of a flock must face/ fly in the same direction,
- Cohesion - in order to form a flock, the individuals must move close to each other.
During the experiment all the robots must adhere to these three rules. Now, clearly, if they only follow these rules, the flock is not going to find a gas leak. So, they added an element of the PSO algorithm. Since the UAV's must adhere to the flocking rules, they decided to previous velocity component and general best location component, so that the direction for each UAV was only based on its own best location.
Although they do mention specific hardware and operating systems in their paper, the conclusion only mentions the results of a simulation. This simulation was successful in finding the source of the gas leak. However, does not seem to be conducted or was unsuccessful in a real 3-dimensional space.
Evaluation We can see that there has been limited research into the implementation of PSO in real robots. In controlled environments or simulations, the idea seems to work. However, in order to be useful in real-life applications a lot still needs to happen. One problem is how to measure the exact location of the robots. As we saw in the first experiment, a faulty location reading, can easily cause inefficiency in finding the gas source. Another thing is that in the experiments, with the exception of a few obstacles, the robots are completely free to move. In order for this technology to be useful, the robots must be able to move around in an entire building, containing multiple rooms and they might have to go through small passages.
Something else to consider is the communication between robots. In order to increase efficiency, the robots should exchange only necessary information. Also, in these experiments, the researchers can easily see where the robots are and verify that they are in the right place. However, if these robots were to be used in gas leak detection in collapsed buildings, the rescue workers cannot see where the robots are. If they could, the robots would not be needed. So, the robots must also communicate their location once they find the light source. Another thing to consider is if the robots should communicate their location more often to the rescue workers. It is possible that a robot gets stuck, maybe their battery runs out, or something else fails. In such a case, it would be nice to know where the robot is, so the robot can be found.
Besides that, it would be very inefficient if all the robots had to be carried out of the building manually. So, while the purpose of the robots is to find the gas leak autonomously, one might consider implementing an override switch to tell the robots to fly back to their basis.
One of the most important requirements of the robobees is the ability of communication between the robots and between the robots and the USAR team. This is because communication plays an important role in the collective perception of the swarm. Collective perception combines the data the individual robots in the swarm have collected into a bigger picture. Cite error: Invalid
<ref> tag; invalid names, e.g. too many
This allows the individual robots to make decisions as a swarm and eventually in our case find the source of a possible gas leak.
Robobees are considered Unmanned Air Vehicles (UAVs) functioning in a swarm. They fly autonomously, and their specific task is to detect if there is a gas leak and report back to the center its location. Both operations require communication. Communication also plays a key role in the coordination of the swarm and control mechanisms Cite error: Invalid
<ref> tag; invalid names, e.g. too many. This communication includes both UAV-to-UAV (U-T-U) communication and UAV-to-Infrastructure (U-T-I) communicationCite error: Invalid
<ref> tag; invalid names, e.g. too manyCite error: Invalid
<ref> tag; invalid names, e.g. too many.
U-T-U communication allows the information to be obtained through sensors. Two UAVs can communicate either directly or indirectly with each other. U-T-I communication is the communication of the UAVs with a fixed control center. This is usually direct communication. For this communication, the choosing of a good routing protocol is important since it plays an important role in achieving reliable end-to-end data transmissionCite error: Invalid
<ref> tag; invalid names, e.g. too many.
What needs to be communicated?
Communciation between robots - In the experiment described above that only uses PSO, they argue, that in order for particle swarm optimization to work, only one value needs to be communicated. That value is the new general best value and its location. The rest of the algorithm can all be computed locally (https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=4223155&casa_token=vTClbprFmjAAAAAA:lXzejZyCbUoXExgKI91RR_SL18LGl3i8SKh3KBwm06f2uzHrynwmOYUZoykhnsRM0L9_dz4D ).
Communication between robot and rescue worker - Of course, the rescue worker does not need to know anything about the algorithm. The only thing that is interesting for the rescue workers is the location of the gas leak. So, we need to find a way for the robots to inform the rescue workers once they have found something. In terms of whát needs to be communicated, the robots need to be able to send their location and possibly the gas concentration value at a certain location to the rescue workers.
Type of information that needs to be communicated - We see that for both the communication channels (between robots and between robot and rescue worker), the only type of information that needs to be communicated is the location of the robots (in the form of a vector) and a simple value of the gas concentration at that position.
How can this be communicated?
When a wireless connection is used to communicate a message between the nodes, there are some restrictions that may interfere with a reliable and continuous communication between the nodes. The main design considerations to look at when using a wireless network compared to using a wired network are: bandwidth of the links connecting the nodes, computing power and energy supply.
When using a routing based network, the number of nodes operating is always lower than using a flooding based network (since the message is conveyed not to all nodes but from one node to another node at a time). This means that a routing based network consumes less energy. However, in modern flooding based networks, less power can be used to achieve the same range since the energy of the signals received from adjacent nodes adds up. For routing techniques, smaller networks consume less energy and for Flooding techniques, larger networks consume less energy (or when the average size of the message is small).
For routing techniques, the range is highly affected by the physical layer and the software of the network. Flooding techniques have a better range since the nodes sum up the energy from all the received nodes.
Latency is not that important in the case of gas leaks, because the timing of the reading is not a deciding factor in the danger of a gas leak. For instance in avoiding a collision, then the time for relaying a message is of high importance because it is a deciding factor in whether the collision will take place or not. In general, routing based networks have higher latency than flooding based networks.
There are different options for communication between the Robobees, and certain requirements to look at. On the one hand we need a fast, reliable and large enough way to communicate and on the other hand we need the Robobees to be able to fly. When we look at different communication methods, we have: physical connection, bluetooth, a LAN either using ethernet or WiFi, or another way to communicate using the environment.
A physical connection means that the different robobees are connected by wires. It does not take long to see that this way of communicating between the robots is impractical. The wires could get in the way when trying to travers difficult terrain and are too heavy to keep the robots flying.
A more promising way of communicating would be Bluetooth. Bluetooth is designed for portable devices using battery power, therefore it already has the conditions needed for swarm technology: low costs, low power and compact size. It allows various types of electronic devices to communicate with each other through a wireless connection, without direct action from an user. When bluetooth devices come in close proximity with each other, they electronically and automatically communicate and establish if they have data to share or when one needs to control the other. Bluetooth devices can only communicate with each other if they have the same profile and capabilities.
Bluetooth consists of four classes. Class 1 is used for long range with a maximum range of approximately 100 meters. Class 2 is the most promising at a range of about 10 meters.Cite error: Invalid
<ref> tag; invalid names, e.g. too many The higher classes have a range of less than a meter which could be too small for our purposes. The range of all these classes are reduced by objects that stand in the way.
Class 2 would probably be the best option for communication between the robobees. When looking at the weight of the class 2 bluetooth transmitter and receiver we get an average of 20 grams which is light enough to be put on the robobees. A small problem with these bluetooth transmitters is that when the source of the gas leak has been found, it could be difficult to transmit this to the USAR workers.
LAN has two different ways to connect. Firstly, we have ethernet which needs a physical connection, so this would not be ideal for Robobees. Secondly, there is WiFi, which needs larger transmitters and receivers to work than a Bluetooth connection, which would add to the weight of the Robobees. However, WiFi could work better for longer distances.
As described above, there are three main technologies for sending messages between the small robots. This concerns Bluetooth, infrared or wireless LAN, both the advantages and disadvantages of the technologies will be discussed after which the most appropriate technology will be recommended. http://www.swarmrobot.org/Communication.html#:~:text=The%20most%20common%20technologies%20for,with%20robots%20not%20in%20sight.
The Infrared technology
- Infrared communication requires the robots to be in direct vision of each other. This can be seen as both an advantage and a disadvantages at the same time. The purpose of using a swarm technology, lays in the possibility of multiple robots operating at the same time while they’re all heading other directions. This makes the requirement of the infrared technology, a disadvantage. On the other hand, in natural swarms, it can be seen that the members often only interact with their direct environment.
- The robots will not accept nor percept messages from robots that are not present in their direct environment.
- Infrared technology does not require much resources, so it can be applied t micro-robots
- Sources of light, like sun-light can disturb the communication between the robots
- Volume of data that needs to be transmitted can be very high, therefore it can be applied to mid-size robot swarms
- When using this technology, the communication can be disturb due to radiation of units.
- Settled between the size and range of wireless LAN and infrared communication
- Provides only the basic conditions, and works with standard protocols, to communicate between the robots.
- Low power consumption, small packages and relative cheap price
According to Seo et all , Bluetooth is the most suitable wireless technology that can be used for the building of swarm technology systems. In conclusion, for the technology aspect of communication between the Robobees the easiest solution would be to transmit between the robots at a Bluetooth class 2 connection. If there is a possibility of a “Queen bee”, a wifi connection or a class 1 Bluetooth connection could be used to transmit the data to the USAR team.
Centralized communication networks
A swarm of robobees is a relatively small swarm, with only one kind of UAV. These UAVs are also small themselves. A centralized communication architecture might be possible for this case, because this is suitable when the size of the UAV swarm and the coverage area is small and the mission relatively simple (finding a gas leak and reporting its location). With a centralized communication architecture, there is a central node to which all UAVs in the swarm are connected. Each UAV has a one-to-one relationship with and directly receives control commands from this central node. A disadvantage is that the U-T-I distance is greater than the U-T-U distance, which causes a delay. (bron zelf) Since the UAVs have high mobility and need to cover the building, this architecture might not be stable enough. Another problem is that once the ground station breaks down, the entire network is disabled. So a centralized communication architecture has the disadvantage of Single Point of Failure (SPOF).
Decentralized communication networks
Since swarms operate at high speeds and need to cover large areas, they frequently connect and disconnect to the network. This is why UAV swarm ad hoc networks are considered a better choice. This eliminates the need of a constant connection to the ground and certain range restrictions: not every UAV will need to have a long-range connection with the infrastructureCite error: Invalid
<ref> tag; invalid names, e.g. too many.
Single-Group swarm Ad hoc Network
A “single-group swarm ad hoc network” is a network where the internal communication of the swarm does not depend on the infrastructure. The communication between the swarm and the infrastructure is a single point link relying on a specific UAV: the gateway UAV (in the case of Robobees, this can be referred to as the ‘queen bee’). This enables the infrastructure to upload and download swarm information. The other UAVs forward data within the swarm and by sharing information, they optimize collaborative control and improve efficiency. The gateway UAV will need two types of receivers: one for communicating within the swarm (to other UAVs) at low power and within short range and one for communication with the ground at high power and within a greater range. This has the advantage that all other UAVs in the swarm will only need to carry low-cost and lightweight short-range receivers, but are still very useful. It will also result in an extension of the communication range of the network. It does however require that the swarm has similar flight patterns.Single-group swarm ad hoc networks are often disregarded because it requires the UAVs to fly in close proximity (and thus be of small size and be the same kind of UAV). Because in the case of small drones detecting a gas leak, which are also all the same kind of UAV, there is no need to form multiple groups, a single-group swarm ad hoc network can be used.
Another design aspect which needs to be considered within the network is an intra-swarm communication network. Three common intra-swarm communication networks are: ring architecture, star architecture and meshed architecture. A ring network is a closed communication loop through a bidirectional connection. Any UAV can be the gateway node, and when a link between two UAVs fails the target UAV can link back through the communication loop. This architecture thus has stability, but lacks scalability. A star network is where the gateway node is in the middle and communicates with the infrastructure as well as the entire UAV swarm. The disadvantage is a Single Point of Failure: if the gateway UAV fails, the entire system is disabled. A meshed network is a combination of the ring and star architectures, where all UAVs in the swarm have the same capabilities. The information stream can be implemented in different forms and any UAV can act as the gateway node. It has the advantages of both networks, which is why this is the best choice for intra-swarm communication.
Choosing the right routing protocol is important for reliable end-to-end data transmission. For the implementation of routing protocols, routing technologies are needed.
There are six common routing technologies for UAV ad hoc networks: store-carry-forward, greedy-forward, path discover, single-path, multi-path and predictive routing. The concept of these routing technologies is briefly explained in the table on the right. The best choice for our case would be predictive routing technology: this predicts the future position of a node by its current position, velocity and direction, and further chooses the next optimal hop node. Since it is applicable in cases where the position of UAVs change rapidly, it would be a good choice for an UAV swarm ad hoc network.
There are three existing classes of routing protocols for UAV communications: topology-based, SI-based and geography/position-based protocols. SI-based routing and geographic/position-based routing are more suitable for UAV networks, since they are suitable for high-dynamic movement and large number of nodes (so a swarm). These categories both consist of multiple routing protocols which might be used for swarm networks. In the table on the left, a comparison of the different Geographic/Position-based and SI-based routing protocols is given. For our case of the bees detecting a gas leak, it is important to look at both ‘location services’ and ‘large overhead’. Location services are an important aspect since the bees need to be able to report back the location of the gas leak. Large overhead is important since the bees are small UAVs, and thus need to be very lightweight which means a small processor is only available. This means the overhead cannot be too large.
Communicate the location with the rescue workers
We have just looked into how the information can best be transmitted from robot to robot or from robot to rescue worker. Now, we want to find a way to translate the information that was sent by the robot to the rescue worker to something that the rescue workers can understand. There are two ways that the location of the gas leak can be communicated to the rescue worker.
- The robots can do their work, and once they found the right location sent that location to the rescue workers.
- The robot can do their work and while doing their work constantly communicate their location to the rescue workers.
The first option is the least amount of work for the rescue workers. They send in the robots and wait until the robots come with new information. However, this option also causes a lot of uncertainty. So, we have decided to go for the second option. While the robots are flying, they sent their location along with the concentration of gas at that position to a main computer. We want to create a program that translates this information into a map. A concentration that is 'normal' gets a green point on the map. A slightly heightened concentration of gas results in a yellow point. And dangerously high concentrations get red coloured points. This way, the rescue workers know where the robots are and get more information about the gas concentrations in the building. This also solves the problem that there might not be any gas leaks in the building. If all the concentrations were not communicated with the rescue workers, the robots might be performing the PSO endlessly or tell the rescue workers that there is a gas leak while that is not true.
We decided to make a user interface, so that the information that is gathered by the robots is communicated with the rescue workers. Before we could make the user interface, we needed to make an implementation of a robot swarm that uses PSO to find a gas leak. In the video on the right, the implementation is visible. Every iteration, the location of the robots is plotted and the colors in the graph represent the values of the gas concentration. The square is the space in which the particles can move. The circle in the middle of the frame is the location where the maximum value is. The black dots are the particles at their current location. The colours green, orange and red correspond to the value of the location. As the video progresses, we see that the robots move towards the circle in the middle, which is indeed the location that we want them to find. Also, you can see that the values of the locations around the circle are red, which means that the value there is high. As we go further away from the maximum, the dots become orange and then green.
The next step was thinking about the design of the user interface. We thought about specific things the user interface should be able to do:
- The rescue worker should be able to click a start button to start the simulation and end the simulation,
- The rescue worker should be able to choose between a map showing the current locations of the robots and a map showing the gas concentration values measured so far,
- In the map with colors, the rescue worker should be able to customize which concentration is indicated with which color,
- It would be great if the rescue workers can choose the number of robots and choose which robots to use,
- There should be an override switch, so that the robots can be sent to a specific location,
- There should be a button to send the robots back,
- Override the PSO algorithm and move the robots around by themselves.
Especially the last 3 bullet points are difficult to implement. So, we will not be able to do this. But it is interesting to consider this in follow up research.
What is the contribution to the USAR team?
The purpose of deploying the Robobees consists of two parts, on the one hand winning time to reach the deadline of 72 hours. On the other hand, the Robobees contribute to the safety of both the rescuers and the possible victims. To know where the gas leak is located, the location of the RoboBees must be communicated to the rescue team. What exactly needs to be communicated and how that has been described before. When the exact location of the gas leak is known, action may be taken according to the protocol. If there is no gas leak, the team can immediately proceed with the remaining steps necessary to explore the condition of the collapsed building. Communication between the RoboBees is not directly relevant to the rescue team. However, this communication affects the time it takes for the RoboBees to find the gas leak. The more the RoboBees have to communicate with each other, the longer it takes for the gas leak to be found effectively. The key to success is therefore efficient communication between the bees so that the location of the gas leak is found as quickly as possible.
Furthermore, as mentioned earlier, the information communicated between the Robobees and the rescue team must be brief and concise. Based on this information, the rescue team will then take further steps to stop the gas leak and bring possible victims still present in the building to safety. It is important to take into account the fact that the gas leak can also be located in a place that is unreachable for the rescuers since the Robobees are operating in a collapsed building that cannot be pre-determined. This is something that can still be investigated in further research.
Week 1 and 2
|Name||Student ID||Time spent||Tasks|
|Jasmijn de Joode||1358073||4.5h||Write deliverables (0.5h), Do research (4h)|
|Robin Foppen||1394746||4.5h||Do research (4h), Write deliverables (0.5h)|
|Mirre Bosma||1266489||4h||Write problem statement (1.5h), Do research (2.5h), Write wiki page (0.5h)|
|Dirk de Leeuw||1358081||4.5h||Problem statement (2h) Research (2.5h)|
|Job van Heumen||1380036||5h||Do research and write state of the art (4.5h), write wiki page (0.5h)|
|Name||Student ID||Time spent||Tasks|
|Jasmijn de Joode||1358073||3.5h||Research legged robots (3h), edit wikipage (0.5h)|
|Robin Foppen||1394746||4h||Drone research and writing (2.5h), update wiki (0.5h), step 2 research (1h)|
|Mirre Bosma||1266489||7h||Research what needs to be done in case of collapse (4.5h) Order wiki page, add short summary begin sections and reference correctly (1h) Do research to what is needed to screen a building and come up with ideas for robots (1.5h)|
|Dirk de Leeuw||1358081||3h||Research Wheeled and tracked robots(3h)|
|Job van Heumen||1380036||5h||Research about the inspection of damaged buildings (5h)|
|Name||Student ID||Time spent||Tasks|
|Jasmijn de Joode||1358073||6.5h||Meeting about requirements (1h), Research on robot swarms/bees (1.5h), editing wiki (legged robots) (1h), write problem statement and deliverables (3h)|
|Robin Foppen||1394746||5.5h||Meeting about requirements (1h), Research on robot bees (1.5h), editing wiki (1h), writing USE section (2h)|
|Mirre Bosma||1266489||5.5h||Research to robot swarms (1.5h) Work on wiki(2h) Meeting about requirements (1h) Research and write PSO (1h)|
|Dirk de Leeuw||1358081||h|
|Job van Heumen||1380036||5h||Meeting about requirements (1h), writing and editing wiki (robo bees) (1.5h), research robot swarms (2.5h)|
|Name||Student ID||Time spent||Tasks|
|Jasmijn de Joode||1358073||h|
|Dirk de Leeuw||1358081||h|
|Job van Heumen||1380036||h|
- How to: Rescue people trapped in a collapsed building - Indonesia. (2009, October 8). Retrieved May 4, 2021, from https://reliefweb.int/report/indonesia/how-rescue-people-trapped-collapsed-building
- Sheu, L.-R., Shih, B.-J., & Chuan-Wei, W. (n.d.). The search and rescue operation in collapsed building caused by earthquakes: a case study. Retrieved from http://www.iaarc.org/publications/fulltext/isarc2000-183_TC1.pdf
- Hutter, M., Gambardella, L., Ijspeert, A., & Chli, M. (2021, March 4). Legged robots. Retrieved May 5, 2021, from https://nccr-robotics.ch/research/rescue-robotics/legged-robots/
- Markoff, J. (2013, 13 juli). Modest Debut of Atlas May Foreshadow Age of ‘Robo Sapiens’. The New York Times. Retrieved May 5, 2021, from https://www.nytimes.com/2013/07/12/science/modest-debut-of-atlas-may-foreshadow-age-of-robo-sapiens.html
- Wikipedia contributors. (2021, 19 april). Atlas (robot). Wikipedia. Retrieved May 5, 2021, from https://en.wikipedia.org/wiki/Atlas_(robot)
- Biswal, P., & Mohanty, P. K. (2020). Development of quadruped walking robots: A review. Ain Shams Engineering Journal. Published. https://doi.org/10.1016/j.asej.2020.11.005
- Glass, S. (2019, January 9). UAV’s to the Rescue—Defining a Purpose-built Drone for Collapsed Building Search & Rescue. Retrieved May 5, 2021, from https://oinkodomeo.com/uavs-to-the-rescue-defining-a-purpose-built-drone-for-collapsed-building-search-rescue/
- Willis, H., Castle, N., Sloss, E., & Bartis, J. (2006). Protecting Emergency Responders, Volume 4: Personal Protective Equipment Guidelines for Structural Collapse Events. Santa Monica, CA; Arlington, VA; Pittsburgh, PA: RAND Corporation. Retrieved May 12, 2021, from http://www.jstor.org/stable/10.7249/mg425niosh
- BBC News - Earthquake rescue: How survivors are found. (n.d.). Retrieved May 4, 2021, from http://news.bbc.co.uk/2/hi/americas/8459653.stm
- J. Tran, A. Ferworn, C. Ribeiro and M. Denko, "Enhancing canine disaster search," 2008 IEEE International Conference on System of Systems Engineering, 2008, pp. 1-5, doi: 10.1109/SYSOSE.2008.4724181.
- J. Tran, A. Ferworn, M. Gerdzhev and D. Ostrom, "Canine Assisted Robot Deployment for Urban Search and Rescue," 2010 IEEE Safety Security and Rescue Robotics, 2010, pp. 1-6, doi: 10.1109/SSRR.2010.5981564.
- RoboBees: Autonomous Flying Microrobots. Retrieved May 19, 2021, from https://wyss.harvard.edu/technology/robobees-autonomous-flying-microrobots/
- Robotic insects make first controlled flight. Retrieved May 19, 2021, from https://wyss.harvard.edu/news/robotic-insects-make-first-controlled-flight/
- RoboBee powered by soft muscles. Retrieved May 19, 2021, from https://wyss.harvard.edu/news/robobee-powered-by-soft-muscles/
- The RoboBee flies solo. Retrieved May 19, 2021, from https://wyss.harvard.edu/news/the-robobee-flies-solo/
- Using static electricity, RoboBees cling to surface. Retrieved May 19, 2021, from https://wyss.harvard.edu/news/using-static-electricity-robobees-cling-to-surface/
- Programmable Robot Swarms. Retrieved May 19, 2021, from https://wyss.harvard.edu/technology/programmable-robot-swarms/
- Communication in Swarm Robotics (slideshare.net)/
- Tahir, A.; Böling, J.; Haghbayan, M.-H.; Toivonen, H.T.; Plosila, J. Swarms of Unmanned Aerial Vehicles—A Survey. J. Ind. Inf. Integr. 2019, 16, 100106.
- TMorse, B.S.; Engh, C.H.; Goodrich, M.A. UAV video coverage quality maps and prioritized indexing for wilderness search and rescue. In Proceedings of the 2010 5th ACM/IEEE International Conference on Human-Robot Interaction (HRI), Osaka, Japan, 2–5 March 2010; IEEE: New York, NY, USA, 2010; pp. 227–234.
- Yanmaz, E.; Costanzo, C.; Bettstetter, C.; Elmenreich, W. A discrete stochastic process for coverage analysis of autonomous UAV networks. In Proceedings of the 2010 IEEE Globecom Workshops, Miami, FL, USA, 6 December 2010; IEEE: New York, NY, USA, 2010; pp. 1777–1782.
- Sahingoz, O.K. Networking Models in Flying Ad-Hoc Networks (FANETs): Concepts and Challenges. J. Intell. Robot. Syst. 2013, 74, 513–527.
- Poli, R., Kennedy, J., & Blackwell, T. (2007). Particle swarm optimization. Swarm Intelligence, 1(1), 33–57. https://doi.org/10.1007/s11721-007-0002-0
- Sempf, A. (2018, 11 april). Swarm Communication - CӔLUS Concept. Medium. https://medium.com/c%D3%95lus-concept/swarm-communication-33cffc47db6d
- Seo, Sang-Wook & Yang, Hyun-Chang & Sim, Kwee-Bo. (2009). Behavior Learning of Swarm Robot System using Bluetooth Network. International Journal of Fuzzy Logic and Intelligent Systems. 9. 10.5391/IJFIS.2009.9.1.010.
- Chen X, Tang J, Lao S. Review of Unmanned Aerial Vehicle Swarm Communication Architectures and Routing Protocols. Applied Sciences. 2020; 10(10):3661. https://doi.org/10.3390/app10103661