# Designing the robot

## Preliminary Designs

From the literature analysis it became evident that robotics technology used for reforestation is still in its infancy, rendering us a plethora of options to design a new robot or improve on an existing model. For this project, we've decided to focus on designing a seeding mechanism for the robot, as this is ultimately the primary functionality of the robot. In this section several preliminary designs have been developed conceptually focusing on different, mostly used options for the seeding mechanism. After the different possibilities of seeding mechanisms are discussed, an informed decision is made as to which seeding mechanism(s) will seriously be considered for actual physical development looking at the different requirements, preferences and constraints that can be drawn from the literature review, case studies and product analysis.

General information about the project can be found over at PRE2017 4 Groep6.

## Preliminary Designs

### Drill

For many seeds, being planted into the ground, at a certain depth, is beneficial for their growth chances. To this extent, a drill would work great. Either the drill could be made hollow, so it could drill a hole and drop the seed instantly, or a separate drill and seed dispenser could be made. By assembling the drill in the middle of the vehicle, and thus most likely with a hole in the base of the vehicle, the most force could be asserted onto the drill. Even though a hollow drill would be the best functioning mechanism, it would be hard to make in practice within the limited time given, which is why two separate mechanisms seem like the obvious choice. This would mean one drill, being assembled at the middle of the vehicle, and one seed dispenser, assembled at the back of the vehicle.

This dispenser could either truly put the seeds in the ground, by for example putting the seed on the end of a stick and pushing this stick into the ground. Or, it could drop the seed into the pre-drilled hole. The latter of these two options would be easier to produce, as the location of the hole can easily be found, (using the relative positions of the drill and dispenser) and it would save an entire part going into the ground, which is deemed a difficult part. The main advantage of the method truly putting the seed in the ground is that it minimizes falling trauma for the seed, as it is gently inserted into the ground. This, however, should not be a problem for the seed, as most seeds are used to being carried by the wind, and thus falling from far bigger heights than the ones talked about here. The main advantage of using a drilling mechanism is also one of its biggest drawbacks. Using a drill, you can very specifically control where the drill is used, and thus the upcoming forest can be planned down to the centimeter (assuming all seeds do sprout). This is great, as it can make sure that all species are there in the desired ratio’s, and everything can be planted as close as possible to the desired location. But it does add the difficulty of navigation. It is very hard for the robot to find out exactly where it is right now, and thus where it should plant. As, when a planning is made for which seed to be placed where, down to the centimeter, the robot should also be able to find its own location, down to the centimeter. This necessary feature for the robot when a drill mechanism is used, is one that is difficult to get functioning precise enough, which is why it is not the focal point of this project. If the option for a drilling robot is chosen, the navigation issue will be left for further research.

An issue that should be dealt with is the one of the drill exerting a lot of force on the vehicle. In order for the drill to truly make a hole, a lot of force needs to be applied. Even if the ground is fairly ‘soft’, the robot is not envisioned to be either very large or very heavy, meaning that the force is big, relative to the robot's size. This means that if the drilling mechanism is chosen, clear attention should be paid to the force it takes to drill this hole, and what the robot's weight needs to be for this not to be a problem.

• A high percentage of seeds planted do sprout successfully
• Not many extra seeds needed
• Precise, provides much control over where seeds are planted
• Seeds planted in the ground are safe from animals/weather

• Drill sticks out when driving, less stable
• Not as fast as other mechanisms
• Relatively big risk that the drill might break/get stuck
A 3D model of an auger

The full model can be viewed here

### Gritter

For rapid deployment of seeds, a gritter like structure could be used to distribute the seeds over large areas. The gritter would give a natural feel to the newly planted seeds since there is no mechanism that determines the position of each seed with great accuracy. This would result in a ‘natural’ feel to the newly created ecosystem. The most important factor of the gritter would be the composition of the seeds since it is random what seed is placed at which position. Multiple storage units could be used, each with a different composition of seeds, to facilitate a variance in compositions that are not totally based upon probability. An example would be a distinction between types of plants. Container 1 could primarily contain grass type plants, while container 2 mostly contains shrubs and bushes and container 3 consists of various types of trees. This creates a multitude of different areas each with a slightly different composition. So in terms of biodiversity, a gritter could place the seeds in such a way that the artificial forest does not show any differences compared to the original forest. The containers would be placed above the gritter and would feed into a funnel to provide the gritter with the desired seeds. Growth enhancers can also be easily added to the mix to help speed the process of creating a new forest. The growth enhancer can be directly added into the mix in the container and no extra process is needed since everything goes through the same gritter from the containers.

However, the gritter system also has serious disadvantages to consider. Since all seeds are basically placed on top of the soil in all cases, seeds may not have a high germination rate for species that require seeds to be position underneath the ground. This can be a direct result from the species itself, as well as external factors. Animals would not have a hard time picking up the seeds from the ground and eating them and weather like rain could potentially wash the seeds away from the area where they are needed. Furthermore, this makes the robot unsuitable for steep hills and areas that have a lot of height differences between them. Rain would wash everything to the lower points within the forest, resulting in a few places that are extremely dense in terms of plants and a lot of places devoid of plants. This also directly affects the resource management of the robot. While the robot could be extremely fast compared to the other preliminary designs, the robot also wastes a lot of resources that could have been used in other areas where they are needed.

In terms of stress on the robot, the gritter requires a low amount of energy compared to the other designs. The system is not directly in contact with the soil and gravity does most of the work to distribute the seeds. The biggest issue would be overall weight of the robot, since a lot of seeds have to be distributed to compensate for the low germination rate. So most of the energy that is needed for this system would go into transporting a large number of seeds in potentially multiple containers. The containers also have to be placed at a higher point in the robot than the gritter itself, meaning that the center of gravity could be relatively high. This could result in an unstable robot, so a robot with a relatively high width and length would be needed to stabilize it.

• Relatively fast compared to other preliminary designs.
• Possibility for high diversity in seeds.
• Easy to add growth enhancers (e.g. compost).
• High variability to keep “natural” looks.

• Seeds vulnerable for animals.
• Seeds vulnerable for weather effects.
• Low hatching rate due to seeds being placed in suboptimal places.
• Low resource efficiency in terms of seeds.
A 3D model of a gritter

The full model can be viewed here

### Plough robot

For seeds to develop properly, they have to be implanted into the soil to receive the necessary nutrients to develop into a tree, therefore some holes or pits need to be present or made. One such possibility for creation holes is by means of plowing, albeit a traditional plow will turn over the soil in one long uniform stretch and hence create some sorta ditch in which seeds can be sown into en masse, usually by means of machinery. Obviously, a traditional plow will not be the desired method for tree planting considering grown trees (the desired end-product of the robot operation) will take up a considerable amount of space. This method will consequently create a uniform and homogenous forest, not only hurting the biodiversity but also completely distorting the natural look of the surrounding area. Therefore, albeit a traditional plow is a very useful tool in the agriculture sector, for the purpose of a reforestation operation after a forest fire in a national park, a variation of the traditional plow will be considered. The envisioned plow will consist of a cylinder which is powered by high acceleration variable torque servos, such that the force applied to the soil can quickly be altered as to create a custom hole for every type of seed. This will result in a complex control system, however from the literature review it is evident that each seed has its own optimal sowing depth, making this a beneficial feature for the seeding mechanism to have. To promote a semi-random seeding pattern the plow will have conjugated pairs of sickles attached to the cylinder which smash the ground to locally create a hole for seeding. The sickles are in conjugated pairs in the sense that they will operate mirrored to each other, i.e. if one blade is about the smash the ground the other will be lifted up in the air. Since we desired a hole to be made locally, the sickles will not be made to perfectly fit the cylinder by cutting out a circle in them as this implies the turning of the cylinder will still create strokes of plowed land which is exactly what we do not want to accomplish. Instead an elliptical hole will be cut out from the sickles such that when they hit the ground and the robot continues to move, the cylinder will just move in the empty space left in the ellipse causing the blade to remain stationary, until the cylinder reaches the back-end of the ellipse, pulling the blade back up from the ground and leaving a hole. This method of plowing, albeit mechanically complex also has a mechanical advantage; as half of the total amount of sickles is suspended in the air during the total operating cycle and the sickles which strike the ground will not be pulled along it which severely reduces frictional forces and hence stresses on the materials of the robot. Because of the elliptical nature of the trajectories of the sickles around the cylinder and the steady turning rate of the cylinder ω a characteristic time τ exists between the events of the first sickle of the conjugated pair striking the ground and the second sickle of the conjugated pair striking the ground. This characteristic time along with the velocity v and the horizontal distance between sickles of a conjugated pair x gives the distance between every successively planted trees $\displaystyle{ r = \sqrt {x^{2} + {(\tau v)}^{2}} }$ . Due to the horizontal distance x between the conjugated pairs of sickles, a linear tree pattern is eliminated as the trees will be planted in a 2D geometry. As the biodiversity requirement needs to be fulfilled, a variable turning rate ω is required to increase the torque and hence the force the sickle exerts on the ground to ensure different depths. However a turning rate increase can be done discretely such that $\displaystyle{ r = \sqrt {x^{2} + {(\tau_{j} v)}^{2}} }$ still holds for the all tree species, albeit the characteristic time $\displaystyle{ \tau_j }$ is now a parameter of the tree species which is being planted j. Using such a plow a pattern of trees can be planted without disrupting an entire haul of the forest floor.

The seeding mechanism can either be made separately from the plow or in affiliation with it. If the seeding mechanism is made separately from the plow the robot is required to make a second run over the terrain to sprinkle the seeds in the holes or a seeding mechanism would have to be made behind the plow. This is because the holes first need to have been dug before the seeds can be planted, and pulling significantly reduces the friction force the robot experiences from plowing than pulling. Although the new design of the plow will probably allow the plow to be placed in front of the robot since pulling drag will be considerably reduced compared to a traditional plow, even though drag forces will still be present to some degree since a perfect sickle cycle will not be achievable. In the latter case, the plow can be put in front of the robot and the seed dispenser in the back of the robot. If the seeding mechanism is made in conjunction with the plowing mechanism, a seed dispenser could release a flow of seeds in between the conjugated single sickles to ensure they fall in the holes that were made just prior to the seed dropping. This method has got some disadvantages as well since allowing the seeds to flow between rotating sickles increases the odds of damaging the seeds and small imperfections in the terrain such as bumps could offset the direction in which the seeds fall on the ground, rendering them vulnerable on the unploughed soil if they fall next to the holes. A third option would be to lift the plowing mechanism up and protrude a seeding mechanism out of the robot. Then, if this process can be made fast enough, this lifting, protruding, seeding and retracting of the seeding mechanism could take placing during the characteristic time τ such that only one run is needed to seed an area. Alternatively, the cylinder could be turned off in this process and resumed once the plow is withdrawn from the robot again. In summary:

• Reduction of the linear pattern of tree planting to maintain a natural look.
• The plow is placed behind the robot, so any holes made will not be affected by the motion of the robot, i.e. if the plow were placed in front the robot might drive over some of the holes which were made just momentarily, which could potentially close or damage the holes and hence the seeds if they are already dispersed
• See the first disadvantage, if the sickle mechanism will be made such that it can be flipped, which implies a modular design approach as the plow module is required to be detachable to make any changes to it, then if one side of the sickles is damaged or become dull the segment can be flipped such that the sickle at the other side can be used. Alternatively, the sickle can be made attachable to the segment spinning around the cylinder, such that a broken part can easily be taken off and a replacement can be inserted. If the robot is made modular it could potentially be reused for other functionalities if the right modules exist.

• Mechanical wear and tear will be an issue for the sickles, since they will be exerting a large force during a small time period to the ground, causing lots of stresses in the material. However, as mentioned in the third advantage by making the sickle segment flippable it could increase the lifetime of the mechanical part twofold.
• Considering the seeding mechanism has either the option of being at the back of the robot at the expensive of the plough being in front, which could cause damage to the holes if the robot drives over them, or being swappable with the plowing mechanism at the back, which would require a longer operation time of the robot to seed a given area, this method is most likely not the most efficient.
• Of all the proposed methods in this section, this will be mechanically the most complex system to design, if not mechanically impossible.
A 3D model of a gritter

The full model can be viewed here

## Conclusion

When looking back at the desired features of a reforestation robot, the robot must be cheaper than current manual reforestation, the robot must be less labor-intensive and must exert a good amount of control in order to restore biodiversity as good as possible. The case study further showed that the robot must be adaptable to work in different contexts, as every National park has different needs. The user analysis exposed the preference of the robot being as harmless as possible for the environment and the robot being easy to use and profitable. On the basis of these preferences, the preliminary designs will be judged and the best design is considered for application in the reforestation robot.

Since this part only focuses on the seeding mechanism of the robot, several preferences are not applicable to this preliminary design. All three seeding mechanisms do not require extra labor and all these three designs thus make sure the work is less labor intensive, the reason why a reforestation robot will be designed in the first place. The analysis will, therefore, focus on the other preferences.

One important preference is the ability to restore biodiversity in a National park. This preference can be considered most important since this is the main goal of the robot. The gritter is the worst design for this preference since it more closely resembles natural reforestation, rather than manual reforestation. When using the gritter, it is not able to exert a lot of control over what seeds will be planted where and also the seeds cannot be planted into the ground, they can only be distributed on the ground as the only changeable parameters of the gritter are the type of seeds that can be dispersed, their relative occupation and seeding rate. This results in several species that are unable to grow and thus is biodiversity not completely restored. Because the gritter has low levels of control and is unable to restore biodiversity this option is ruled out of being suitable for this situation. The drill and the plow are, however, able to exert a decent amount of control in order to restore biodiversity. Therefore, the focus will from now on be on the drill and the plow.

Both the drill and the plow are able to restore biodiversity because both mechanisms are able to plant different seeds at their ideal depth which optimizes the survival rate of all species. This enlarges the opportunity for full recovery of the biodiversity, as is explained in in the Extended Literature Review. One main advantage of the drill over the plow is that the drill is mechanically less complicated to produce. This not only means that it is easier to produce but has the extra benefit that it is therefore more profitable for companies to develop. This is an important factor since if there are no companies wanting to invest in the reforestation robot, no reforestation robot will be developed. When looking at the efficiency at which these methods are able to restore biodiversity, the drill is approximately four times slower than a plow, however, the survival rate of seeds when planted with a drill is higher than when planted with the plow (Preece, Oosterzee and Lawes, 2013)[1]. Since it is deemed more important to restore biodiversity in an efficient way rather than seeding as swift as possible, the drill mechanism is considered the best mechanism to restore biodiversity.

Additionally, the drill design has more inherent freedom than the plow design, this is because the plow is geometrically constrained to be at the back end of the robot, to reduce frictional forces as earlier explained. This means that the drilling mechanism can be incorporated anywhere on the robot. This gives the extra of building a small scoop at the back end of the drilling robot, which will serve to use some of the top layer soil to plug the hole made by the drilling robot, such the seed is truly buried and protected.

Further preferences that have not yet been discussed are that the robot is harmless for the environment and the required versatility of the robot to be able to satisfy the variable needs for different contexts. Both the drill and the plow have the same qualities regarding these preferences and there is no real preference for one mechanism over the other based on these two parameters.

Concluding, the drill satisfies most preferences of the design. The mechanism is able to restore biodiversity by exerting a good level of control, promotes the highest survival rates for the newly planted seeds and has the highest probability of being profitable for companies. Therefore, this mechanism is chosen as the seeding mechanism for the final design of the reforestation robot.

## Bibliography

1. Preece, N. D., Oosterzee, P., & Lawes, M. J. (2013). Planting methods matter for cost-effective rainforest restoration. Ecological Management & Restoration, 14(1), 63-66. https://doi.org/10.1111/emr.12017