This paper presents a general method for planning collision-free wholebody walking motions for humanoid robots. First, we present a randomized algorithm for constrained motion planning, that is used to generate collision-free statically balanced paths solving manipulation tasks. Then, we show that dynamic walking makes humanoid robots small-space controllable. Such a property allows to easily transform collision-free statically balanced paths into collision-free dynamically balanced trajectories. It leads to a sound algorithm which has been applied and evaluated on several problems where whole-body planning and walk are needed, and the results have been validated on a real HRP-2 robot. IntroductionDuring the last twenty years, impressive progress has been achieved in humanoid robot hardware and control. This leads to a rising need for software and algorithms improving the usability and autonomy of those robots. One important area of research focuses on the development of robust and general motion generation techniques for safe and autonomous operation in human environments, such as offices or homes.Motion planning for humanoid robots is challenging for several reasons. First, the computational complexity of classic motion planning algorithms is exponential in the number of Degrees of Freedom (DoFs) of the considered system, which is high for humanoid kinematic trees. Second, a humanoid robot is an under-actuated system: the DoFs that control the position and orientation * The authors are with CNRS, LAAS,

Abstract-This paper deals with motion planning for a humanoid robot under task constraints. It presents a novel random method, inspired by the RRT-Connect algorithm, that uses a local task solver to generate statically stable collision-free configurations. It is able to plan motions for a large variety of tasks. In an experimental section, we compare in details this random strategy with a local collision avoidance method found in the literature.