Taylor Howell

CYBERDRIFT

This is a motion-planning experiment to perform a parallel park with an autonomous vehicle. The parking spot is so narrow that it requires the vehicle to drift into the goal position. Contact-implicit trajectory optimization is used to optimize a path with initial and goal states, as well as the vehicle dynamics and contact interactions. The vehicle is modeled as a simple wheeled vehicle with particle z-dynamics and a single-patch friction cone that produces maximum-dissipation Coulomb friction.

CYBERJUMP

The same model is now used to perform a ramp jump with the autonomous vehicle.

CYBERTOSS

In this experiment, the objective is to move an object to a goal position using a simple manipulator. Contact-implicit trajectory optimization is used plan behavior that initiates contact between the end effector and object in order to move the object. The manipulator is modeled as a double pendulum and the object as a planar particle with a single-patch friction cone that produces maximum-dissipation Coulomb friction.