7. Integration of the Whole-Body Controller into a Higher-Level Framework

This chapter serves as an outlook to prospective challenges in robotics, where compliant whole-body control will act jointly with an Artificial Intelligence (AI). The field of service robotics, for example, puts high requirements on the systems due to the complexity of household chores: The environment is usually dynamic and unstructured, and a wide variety of tools with different contact properties exists. These aspects require an elaborate task planning, both from a logical and a geometric perspective. Moreover, the close cooperation of all modules in a robotic system is necessitated: A whole-body controller for soft physical contacts requires a proper parameterization, i.e. controller gains, a specified control task hierarchy, trajectories and goals, to perform the tasks. A non-deterministic, AI-based planner can provide these data while not necessarily being hard-real-time-capable itself. In case of local minima on the control level, the planner is able to reschedule to find feasible, global solutions. Other modules such as the vision system or the speech recognition may also be triggered.