Скачать или смотреть Anytime Whole-Body Planning/Replanning for Humanoid Robots

We consider the problem of generating whole-body motions for a humanoid that must execute a loco-manipulation task, i.e., a manipulation task that implicitly requires stepping (e.g., reach a desired set-point with one hand). Furthermore, the robot must plan an initial partial solution within a specified time budget, after which it has to start moving.


To solve this problem we propose an anytime planning/replanning algorithm that interleaves planning and execution intervals: a previously planned whole-body motion is executed while simultaneously planning a new solution for the subsequent execution interval.
At each planning interval, a randomized local planner builds a tree in configuration-time space by concatenating successions of CoM movement primitives. Each primitive represents an elementary humanoid motion (such as static/dynamic walking, jumping and squatting). The planner works in two stages. A first lazy stage quickly expands the tree, testing only vertexes for collisions; then, a second validation stage searches the tree for feasible, collision-free whole-body motions realizing a solution to be executed during the next planning interval.


This video clip explains the working principle of the proposed framework and reports planning experiments performed in V-REP for the small-size humanoid robot NAO. In particular, we show results for three different scenarios of increasing complexity.