We designed a robotic system to play baseball in Drake using trajectory optimization.
Baseball is a task that contains two interesting manipulation paradigms: Pitching & Hitting. There is limited literature on nonprehensile manipulation for dynamic aerial objects. We approach this using planning techniques including inverse kinematics & kinematic trajectory optimization.
This project was done for 6.4200 (Robotic Manipulation) & the code can be found here. The dependencies can be installed with Docker.
Methods
Some key results of our approach after testing are below:
View Project Presentation (PDF)