Planning end effector trajectories for a serially linked, floating-base robot with changing support polygon

We present a framework for the planning of end effector trajectories for serially linked robots with a non-fixed, also known as floating, support base. The FBSR (Floating Base, Serial Reach) planning algorithm extends the capabilities of serially linked, floating-base robots by planning to maintain stability while transferring modules from a weight-bearing role to the active, manipulator-like portion of the robot as needed to extend the robot's reach. During planning, we employ several methods such as random restarts and dynamic weighting of the optimization cost function sub-goals to reliably provide stable solution trajectories. We experimentally validate this algorithm on a 16 degree of freedom snake robot, and demonstrate the successful execution of generated plans.