Samenvatting

In this graduation project, GreenTech Labs want to create a crop-weed-removal robot-arm. However, they want to achieve this using modern tools, with a system that can help them
automate and simplify the process of training custom robot arms like this. The student who was assigned to this project had to research into the hows and whys of this situation, and
propose and build a design that lets GTL train custom robot arms using Robot Operating System 2 and various simulation tools.
After much research, trial and error, a system was designed that works in a contained way – it requires the setting of a small amount of hyper parameters and 1 robot-arm model, simulated
in a program called Gazebo – and uses these things to create neural networks from the simulation that the real-life robot would use to perform its tasks, do path finding, etc.
The project is only halfway finished, with the system being able to map out all the areas of movement that a robot can do without bumping into itself. There is more work to be done after handing off is complete, but the project itself serves as a solid proof of concept. Further real-life testing will still be carried out after this report, as it is still not fully done.