“FPGA acceleration of motion planning algorithms for robotics applications”
John Keszler, S.B. ’19, electrical engineering
Advisor: Scott Kuindersma, Assistant Professor of Engineering and Computer Science
As robots are developed for a broad spectrum of uses, from health care to search-and-rescue to personal assistance, there is a growing need to develop fast, real-time motion planning algorithms to enable robots to be integrated more seamlessly into these real-world applications. Keszler built a hardware controller for the motion or trajectory planning of a three-link robotic arm, and then used it to test the feasibility and correctness of the motion-planning algorithm he developed. Since many hardware solutions are computationally limited, Keszler hopes that, by working at the intersection of hardware and software, his project could lead to a potential solution to a common, but computationally difficult, problem. He said the biggest challenge involved data synchronization and transmission, since he was dealing with data transfer at a very high speed between hardware and software systems.
“While this project is looking at a specific problem for a specific demo, the algorithm that I chose can be generalized to nearly any robotics platform,” he said. “And as I have found out in my project, the potential for parallelism and improve performance of my accelerator remains consistent, if not better, with more complicated problems.”