Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Trajectory Optimization for Quadrupedal Locomotion
A challenging problem in legged robotics is to generate motion plans that are feasible with respect to the robot’s skills (range of motion, actuator limits) and the underlying physics (dynamics, contact constraints). State-of-the-art approaches tackl
A challenging problem in legged robotics is to generate motion plans that are feasible with respect to the robot’s skills (range of motion, actuator limits) and the underlying physics (dynamics, contact constraints). State-of-the-art approaches tackle this highly nonlinear and non-smooth optimal control problem by transcribing it to a finite horizon problem with discretized state and action space and solving it with modern nonlinear programming methods. The key to success of those methods strongly depends on the way the constraints and objectives are formulated. Mordatch et al. presented a method to deal with the contact constraints to synthesize various behaviors for virtual characters as shown in the video: https://youtu.be/mhr_jtQrhVA. The task of this project is to implement and extend their promising approach for our quadruped robot.
A challenging problem in legged robotics is to generate motion plans that are feasible with respect to the robot’s skills (range of motion, actuator limits) and the underlying physics (dynamics, contact constraints). State-of-the-art approaches tackle this highly nonlinear and non-smooth optimal control problem by transcribing it to a finite horizon problem with discretized state and action space and solving it with modern nonlinear programming methods. The key to success of those methods strongly depends on the way the constraints and objectives are formulated. Mordatch et al. presented a method to deal with the contact constraints to synthesize various behaviors for virtual characters as shown in the video: https://youtu.be/mhr_jtQrhVA. The task of this project is to implement and extend their promising approach for our quadruped robot.