Towards integrated walking and jumping motion planning in complex environments: Jumping trajectory generation

This paper presents the initial research for a framework by which a biped robot can make a navigation plan in a obstacle filled environment by performing both walking and jumping motions. In particular this paper focuses on automatically creating a jumping trajectory based on the distance that the robot is required to jump. This jumping trajectory considers angular momentum so as to reduce backwards jumping and spinning of the base in the air, this is done with the Eulerian ZMP resolution method(EZR). Additionally this trajectory also considers compliance at the landing foot to avoid manipulator damage while ensuring that the robots support foot is extended just the right distance to let a inverted pendulum model with non constant height reach the top with zero kinetic energy remaining. The path planner is based on the A-Star path planning algorithm. The navigation plan also considers finding a path which is sufficiently straight prior to the moment of jumping so that the robot can build up the required linear momentum to execute the jump. The ultimate goal is to create a framework which can consider path planning with various types of motion.