Abstract
In this thesis, we investigate real-time whole-body control of humanoid robots under multi-contact-modes settings. That is to say, under different contact conditions such as a mix of desired fixed contacts and desired (i.e., controlled) sliding ones. Several methods are proposed to maintain the robot's balance and keep the center-of-mass (CoM) within an admissible set. Some of these methods require expensive computations of geometric balance regions. Consequently, the online realization of the balance criteria in multi-contact has been a long-standing challenge. In this thesis, we tackle this challenge in three main steps. First, we present a fast-computing method for the 2D CoM support region in a configuration of multiple fixed and intentionally sliding contacts. To select the most appropriate CoM position within this region, we account for (i) constraints of multiple fixed and sliding contacts, (ii) desired wrench distribution for contacts, and (iii) desired CoM position (eventually dictated by other tasks). These are formulated as quadratic programming (QP) optimization problems. This stage contains computational limitations and can not cover all feasible robot configurations during the scenarios.Next, we propose a whole-body control strategy for humanoid robots in multi-contact settings that enables switching between fixed and sliding contacts under active balance at will.This approach computes a safe center-of-mass position and wrench distribution of the contact points based on the Chebyshev center in real-time and without any computational limitations. Moreover, this region-free approach does not need the geometric computation of balance regions and emph{a priori} computation of them. We assess our policy with experiments highlighting switches between fixed and sliding contact modes in multi-contact configurations. A humanoid robot exhibits such contact interchanges from fully-fixed to multi-sliding and also shuffling of the foot.The scenarios represent the execution of our control scheme in realizing the desired forces, CoM position attractor, and planned trajectories while actively maintaining balance.Finally, we introduce a unified framework for the whole-body dynamic balance controller of humanoid robots in multi-contact as an alternative to controlling the balance in a separate thread (planning). This framework considers the active motion tasks of the robot in real-time within the balance criteria of the robot. We illustrate the applicability of each step by simulations and empirical experiments on the HRP-4 humanoid robot.