Simulation of a robotic prosthesis: Using reinforcement learning
Robotic prosthesis is becoming more and more popular today. Robots trained to interact with objects can be used in different fields as it opens the possibility of having robots that interact more smoothly with the environment. In recent years, there have been numerous breakthroughs in using reinforcement learning to program robots. This has resulted in robots being able to solve tasks dynamically and reasonably. Aiming at making robot agents eventually act sensibly when interacting with the objects, we explore two different kinds of programming techniques, using reinforcement learning, and by programming the robot to work based on sensor input. The experiment results demonstrate that the agent acting based on sensor input performed best with limited resources, making a robotic arm operating using sensors a conceivable prosthesis.
Undergraduate thesis submitted to the Department of Computer Science, Ashesi University, in partial fulfillment of Bachelor of Science degree in / Computer Science, May 2020
robots, reinforcement learning, simulations, programming techniques