Dynamic Obstacle Avoidance on a Self-balancing Robot Platform

Autonomous tasks are increasingly becoming part of our everyday life, whether in a factory floor where robotic manipulators manufacture goods, or when cars acquire the capability of parking themselves. These tasks often involve a human operator that has some high level control over the system. This can lead to situations where the human operator believes that the system is safe, when this might not be the case. This is why robust control algorithms need to be implemented that provide safety guarantees when a mechanical device is teleoperated by a human user. These algorithms can intervene when an unsafe choice is made to protect the operator and the system itself. The challenge that this work specifically focuses upon is dynamic obstacle avoidance by a robotic unit that is guided by a human user. Dynamic obstacles are objects that move in the environment independently from the robot and can potentially raise safety concerns for the robotic platform. In order to detect these obstacles in the environment, sensor data along with some probabilities need to be utilized to infer the magnitude and direction of an object’s speed. This project is primarily concerned with this estimation and methods to do this