Agimus: a new framework for mapping manipulation motion plans to sequences of hierarchical task-based controllers

Abstract

In this paper we present the integration of a manipulation motion planner involving multiple contacts with an automated generator of controllers to execute the manipulation tasks. The novelty of the method is not only to produce a configuration space trajectory but also automatically formulate the controllers that perform the tasks while keeping balance on a humanoid robot. We demonstrate this approach fully integrated on a real Talos humanoid robot while using controllers formulated as a sequence of hierarchical Stack-of-Tasks.