Optimization‐based Full Body Control for the DARPA Robotics Challenge

Abstract

We describe our full body humanoid control approach developed for the simulation phase of the DARPA Robotics Challenge (DRC), as well as the modifications made for the DARPA Robotics Challenge Trials. We worked with the Boston Dynamics Atlas robot. Our approach was initially targeted at walking, and it consisted of two levels of optimization: a high-level trajectory optimizer that reasons about center of mass and swing foot trajectories, and a low-level controller that tracks those trajectories by solving floating base full body inverse dynamics using quadratic programming. This controller is capable of walking on rough terrain, and it also achieves long footsteps, fast walking speeds, and heel-strike and toe-off in simulation. During development of these and other whole body tasks on the physical robot, we introduced an additional optimization component in the low-level controller, namely an inverse kinematics controller. Modeling and torque measurement errors and hardware features of the Atlas robot led us to this three-part approach, which was applied to three tasks in the DRC Trials in December 2013. C 2014 Wiley Periodicals, Inc.

On Weight-Prioritized Multitask Control of Humanoid Robots

“…However all these formulations can be fit in the general framework of (120) with, as such, differences in the particular formulation of the X constraint set and in the choices of the ||τ k −τ [42]. The WPI-CMU team used an equivalent formulation to the one we presented here [43,Eq. (1) and Sec. 5].…”

Modeling Robot Geometries Like Molecules, Application to Fast Multicontact Posture Planning for Humanoids

“…For cyclic motions like walking, model-based approaches dominantly rely on simplifications and periodicity analysis techniques to stabilize the motion [1]. In a larger scale, however, especially for locomotion in complex environments, one would require a plan ahead of time.…”

“…One popular approach is to find a time evolution of internal coordinates that converges to the optimal posture. This can be done via inverse kinematics (IK) [3] or inverse-dynamics (ID) formulations [1], [4] which offer compliance as well. In these time-integration controllers, an optimal posture is found naturally, but with certain dynamics.…”

Model predictive control of biped walking with bounded uncertainties

“…These sources of uncertainty can always be reduced with more precise (and more expensive) hardware or we can try to estimate them, but only up to a point: there always remains a certain amount of uncertainty. And as a result, there always remains a certain amount of tracking error when trying to follow a reference motion [5], [12], [9], [25], [15]. In order to satisfy all constraints in the presence of such tracking error, the usual approach is to introduce hand tuned safety margins [24].…”