The objective of this project was to build a statically stable walking and turning interface for JPL’s Robosimian quadrupedal robot to allow operators to move the robot in real time. This was achieved by building a forward and sideways gaits, as well as a reset procedure. All the motions are statically stable. The software was built using Klamp’t for simulation, inverse kinematics, and physical modeling. The work was done under Kris Hausers’ Intelligent Motion Laboratory at Duke University. Supporting code is available upon request.
Top right simulation is where motion planning is performed. The top left simulation visualizes the simulated physical world. The bottom video shows the robot in motion following the specified motion.