GSoC’22 RoboComp project: Model Predictive Control for obstacle avoidance

30th June 2022

About me

I am a MS by research student associated with the Robotics Research Center in International Institute of Information Technology, Hyderabad. My work is focused on trajectory optimization for autonomous vehicles using Model Predictive Controls.

About the Project …

Model Predictive Control is a form of Optimal Control that can be used to compute robot navigation control actions in real-time, directly from the dynamic model of the robot, constraints on its parameters and LIDAR data. The goal of this project is to first evaluate the performance of the existing MPC and then improve upon the shortcomings. The current implementation uses the free balls algorithm to avoid obstacles. The project will focus on where this algorithm is failing and how to improve upon that. I will also be looking at how to represent obstacles in a better way so that the performance of the MPC algorithm is faster and the solutions are better quality.

Kaustab Pal