Trovador: Tree Planting Robot
The Gap
Deforestation, caused by agricultural expansion, wood extraction, wildfires, and infrastructure expansion is one of the leading causes of global warming. According to a study published in the journal Science in 2018, the loss of forests and other vegetation accounts for about 10% of global greenhouse gas emissions.
There’s simply no way to can fight climate change if we don’t stop deforestation.
Trees are vital, they give us oxygen, store carbon, stabilize the soil and give home to the world’s wildlife. For the health of our planet and its inhabitants, their preservation and protection should be a priority for all of us.
The United Nations set the 17 famous Sustainable Goals to ensure the well-being of all Earth’s inhabitants. One of them, goal 15, aims to preserve life on land. Two of the targets aim to end deforestation and desertification and restore degraded land and forests.
To tackle reforestation, there are a lot of human volunteers and projected machinery to restore the so missed green landscapes. But the targets haven’t yet been achieved. The 2023 Tree-Planting Statistics point out that in the whole world, an average of 483 trees are being cut per second, but only 60 are being planted.
The main reason why reforestation is not being achieved, except for bad forest management, is because human labor brings many challenges in terms of injuries, maintenance, and planting capability.
Another factor is that 30% of Earth's surface is compounded by rough terrain, mountains, and cliffs. It's completely risky to send humans there, and impossible to navigate with machinery.
What is being done is not enough!
On the countdown, I build Trovador
A tree-planting Robot with the capacity of accelerating the reforestation process and restore green ecosystems!
Trovador is a hexapod robot that plants newborn trees, the length of a bottle, and comes as a response to replace human volunteering and machinery work on reforestation.
Structure
Trovador has a similar morfology of a regular hexapod. This limbed robot has several advantages in its locomotion and use as higher mobility since it can move over rough terrain and overcome obstacles way easier than others, higher versatility because its multiple legs can be used for a variety of tasks, like walking, climbing, and digging. Finally, but super crucial, it can operate with higher stability since it maintains its stability on rough terrain and even when faced with unexpected obstacles. This kind of robot is today known as the most efficient because more than 50% of the Earth’s surface cannot be navigated using wheels or tracks alone.
It has six legs each with 3 joints, adding up to 18 degrees of freedom, being these numbers the most efficient. The fact that the hexapod has 6 legs optimizes the locomotion in terms of: Stability, offers a stable base due to their evenly distributed legs. Redundancy, in case of leg failure or damage it can still maintain stability and continue functioning with the remaining legs. Simplified control, with fewer legs than, for example, an octopod (eight-legged) configuration, hexapods can have simpler control systems. Maneuverability, the placement of the legs allows for versatile movement, including turning, climbing, and navigating obstacles.
I built a prototype on a smaller scale and capacity, with recycled materials, just to test out if the idea was possible.
End-effectors
Each leg has a crucial function in the robot's performance.
The middle legs are incorporated with a gripper holder on the tibia, to pick up and plant trees. This gripper was designed to have a quadrangular pyramid as the holder to be able to grab and transport the trees safely. This gripper works with the rotation of a stepper motor that not only serves to open the holder as to do it precisely. The fact that the middle legs are two tree holders/planters it optimizes the power consumption and the materials costs, allowing both walking and planting.
The back legs have a foot on the tip that makes pressure on the ground. This foot is rotative so it can adapt to the terrain and allow stability to the tip.
And the front legs, which only serve as support, have a sharp tip to add grip to the locomotion.
The Body
These end-effectors connect with the body, where all the microprocessors are. The body was designed to have 4 layers:
The top layer has the tree holders, each with its separate space and proper room to allow the gripper to get inside to hold them. It is placed on the top of the robot to allow the gripper to have enough maneuver to grab them. This holder prototype can fit up to 6 trees. (It is literally an egg box :) )
Under the top layer, there is a sub-layer, exactly placed in the middle of the body with the MPU6050 sensor. This is an accelerometer-gyroscope module with a six-axis to measure all kinds of displacement parameters, which has to be in the middle to be the most sensitive to its center of mass.
The second layer is where the microprocessors are placed at. Here I used an Arduino Mega board and an Arduino Uno board in parallel. This is because Arduino Mega uses the ATmega2560 processor, with a flash memory of 256 kb and SRAM of 8kb, whereas the Arduino Uno has the process of type ATmega328P with a flash memory of 32kb and SRAM of 2kb, none of those are enough separately to process all the data run. In the versus of this layer is where all batteries are placed where the tension of 7V and current of 3mAp are run.
The bottom layer serves as protection against the bushes’ friction and dust. It also holds the rotation of the servo motors to ensure the servo rotates exactly vertically.
Planting Loop
Trovador works on optimized planting sequences:
1st- Walk
These sequences begin with a 5 m walking pattern, using a Tripod gate. This means it walks 5 meters in front with the suspension of three legs at a time in order to move more efficiently. This efficiency is because of:
- Tripod Stance: In a tripod gait, the robot always maintains three legs in contact with the ground while the other three legs are lifted and swinging forward. This tripod stance creates a stable triangular support base. As long as at least three legs are in contact with the ground, the robot can maintain stability.
- Even Distribution of Forces: With three legs in contact with the ground, the forces exerted by the robot’s body weight and other external factors, such as terrain irregularities, are evenly distributed among the supporting legs, which is crucial for outside mobility.
- Redundancy: In case of leg failure or damage, the robot can still maintain stability and continue moving using the remaining two legs. This redundancy enhances the robot’s robustness and ability to recover from leg failures.
In this article, I explain on a deeper level this type of locomotion, specifically what are tripod gate and leg cycles: Link
2nd-Tree Picking-Digging
Then, the robot takes a tree from the holders and sticks its gripper on the ground. Since the robot legs were built with servo motors, which are revolute joints, meaning each joint rotates on one axis, the middle legs can turn up and grab the trees from its back, and turn down to plant them in the ground.
3rd-Soil Measurement
Then, the implemented humidity sensor on the gripper measures the ph levels of the soil and determines whether it leaves the tree there or not. If the ph level isn’t high enough (pH > 6), the robot walks 2,50 m forward and repeats the same pH measuring process. And does this until the humidity of the soil is nutritious enough to grow a tree on.
4th- Planting
Because this process is automated, the tree is planted at the right depth, to make sure the root collar is on the same level as the soil.
For optimal health and growth, most trees do best with a soil pH of 6.0–7.0. So, if the soil moisture sensor measures a pH higher than 6 the gripper leaves the tree there and starts to walk 5m again to repeat the sequence. In this next sequence, the robot repeats everything, but now with the gripper of the other side.
5th-Stifling
Whenever the tree is planted, and the robot moves forward, the back leg makes pressure near the root with its rotative foot, to ensure the roots are in tight touch with the soil. This is possible because the hexapod is walking on a tripod gate, and when the leg raises itself from the initial position to the final position on the leg cycle it steps right on the initial position of the leg in front, the one with the gripper.
In parallel
The robot maintains its stability throughout the entire sequence by utilizing a PID controller. This controller processes the provided data from the MPU6050 and generates corrective actions to ensure balance. This controller is comprised of three components — Proportional, Integral, and Derivative — the PID controller calculates the error between the current orientation and the desired orientation.
By employing inverse kinematics, the controller determines the optimal leg rotation required, using the displacement parameters measured from the MPU6050 sensor, to have the robot’s center of mass stable. This stability is approached on both static stability and dynamic stability. These are achieved with those inverse kinematics calculations, which I explain exactly how it works in this article: Link
This is crucial because on the outside there are many obstacles to face, and, while caring a numerous amount of trees, it is necessary to have the center of mass stable.
The future of landscapes
Measuring the distance between those trees planted on both gripper sides, they are distanced on a 5m in length and 1 m in width. The result hypotenuse is ~=5 meters, equal to the minimum distance to plant two trees, optimizing the planting process. That is why every planting sequence begins with a 5m walking pattern. This allows the optimization of the space to accommodate their branching without overcrowding and gives them space for light and air movement.
The planting season must be during spring and autumn, depending on the specie, but these are the best seasons to plant trees since they are the less extreme.
With all these conditions, (ph levels, planting season, planting depth, and planting distance) it is possible to secure a high survival rate for all trees planted.
Trovador’s next steps
The next step of Trovador will be the implementation of reinforcement learning. This is a machine learning training method based on rewarding desired behaviors to improve the tree planting sequence and, most importantly, to replace the balancing controller.
And another huge improvement will be the implementation of a Real Sense Depth camera, that uses stereovision to calculate the depth and overcome the surrounding obstacles more easily.
The version I built is an alpha version, due to the resources I have. This prototype was built with a PVC board, cut, and assembled by hand. Besides the PVC, Trovador is entirely built with recycled materials such as egg boxes and curtain bolts, and other old robot parts. The electronics can be quite expensive to build, due to their complexity and numerous parts. That is also why anything beyond this number of legs and joints, the complexity and hardware costs outweigh the aforementioned advantages.
Although very versatile, this kind of robot has obviously its disadvantages. The robot’s hardware has to work in a very precise way, making its design complex to build by hand. Since the robot was mainly built with recycled materials, its maintenance is quite weak. In addition, it consumes huge amounts of power in little time since it requires power to move each leg, meaning each joint, the robot gets limited endurance and range reducing its working time significantly.
The Trovador Alpha is almost four times the size of a computer. But, to achieve the challenge I am running to, its ideal size will be four times its current size to fit around 300 trees, and plant them in one HOUR!
Trovador Alpha was able to plant a tree in 26 seconds (meaning on a single planting sequence). This equals to planting around 128 trees per hour, on medium-ideal terrain conditions.
On a more personal note:
Trovador comes as a response to accelerate reforestation and keep up with the deforestation rhythm. It is not a solution to deforestation but a solution to overcome the reforestation obstacles that exist on the earth's surface.
What is new? So far there isn’t such a hexapod application. There are drones, and quadruped robots designed to perform similar tasks. But their complexity overcomes Trovador’s, but their efficiency DOESN’T.
Trovador was designed and built to be an accelerator, meaning all the planting sequences and conditions were planned out to build forests that will require minimal care and provide maximum growth.
Author’s homage:
I’m from Portugal and I love my home. In the last 20 years, Portugal has lost around 388000 hectares of green area due to wildfires. It is really concerning watching forests go from alive green to grey in seconds, and watching them go from grey to brown so slowly… I built this robot as an urge to replant my country and a dream to die seeing all Portuguese green landscapes again.
D. Dinis, King of Portugal, also known as The Troubadour King, “Trovador” in Portuguese. Was a visionary that ran Portugal into impressive progressions in the economy, literature, and education. D. Dinis is mostly remembered as the one who demanded to plant Leiria’s Forest, one of Portugal’s greenest lungs.
