Whole-Body End-Pose Planning for High-Degree-of-Freedom Robots on Uneven and Inclined Surfaces

Abstract

During the last few years there have been significant improvements in the field of humanoid robotics. More powerful workstations capable of running more accurate - and therefore more computationally demanding - simulations, and the rise of new generations of humanoid robots with better hardware, have enabled researchers to keep pushing the boundaries and create novel methods to improve the perception and motion of these robots. Motion planning is the area of robotics which concerns with how and when a robot should move a part of itself, and the execution of such motion. Motion planning has been a thoroughly investigated area, but not all of the challenges related to it are solved yet. Robots with a fixed base and few degrees-of-freedom (DoF), e.g. the industrial robotic arms that revolutionized the automotive industry, have been used as a means to approach the problem of motion planning. Often these type of robots are associated with an isolated environment, in which they do not have to interact with people. Researchers have developed successful motion planning algorithms to operate robots in these environments. Nonetheless, those approaches fall short when humanoid robots are taken into consideration.Applications aimed towards humanoid robots have to take into account the characteristics often associated with them: many DoF, a floating base, and balance and dynamic constraints. Implementing autonomous solutions with safe human interaction in complex and dynamic environments, considering biped balance and possible external interferences is non-trivial. Our goal is to tackle the problem of high dimensional kinematic and dynamic motion planning.Namely, we will focus on the sub-problem of humanoid end-pose planning on uneven terrains.