Multi-Contact Heavy Object Pushing With a Centaur-Type Humanoid Robot: Planning and Control for a Real Demonstrator
Matteo Parigi Polverini, Arturo Laurenzi, Enrico Mingo Hoffman, Francesco Ruscelli, Nikos G. Tsagarakis
Abstract
Performing a demanding manipulation task with a multi-legged loco-manipulation platform may require the exploitation of multiple external contacts with different environment surfaces for counteracting the manipulation forces. This is the case of the pushing task of a heavy object, where the grip forces at ground may not be adequate and establishing leg contacts against a wall turns out to be an effective solution to the manipulation problem. In order to produce such behaviour, this letter presents a control architecture that is able to freely exploit the environment complexity to perform loco-manipulation actions, e.g. pushing a heavy object, while meeting the implementation requirements to achieve a real demonstrator. The proposed approach, conceived for torque-controlled platforms, combines the planning capabilities of nonlinear optimization over the robot centroidal statics, based on a continuous description of the environment through superquadric functions, with the instantaneous capabilities of hierarchical inverse kinematics and reactive contact force distribution. Experimental validation has been performed on the pushing task of a wooden cabinet loaded with bricks, using the CENTAURO robot developed at Istituto Italiano di Tecnologia (IIT).