Humanoid Control Workshop - Videos and Slides
Presented at the Institute for Human and Machine Cognition (IHMC) in Pensacola, Florida on March 14, 2014.
Siyuan Feng, CMU
Optimization-based full body control for the DARPA Robotics Challenge
This talk presents the full body humanoid control approach we developed for the simulation phase of the DARPA Robotics Challenge, and the modifications we made for the DARPA Robotics Challenge Trials.
Sylvain Bertrand, Tingfan Wu, Jerry Pratt, IHMC
Momentum-based control framework and capturibility-based walking control – Application to the humanoid robot Atlas
We’ll present our new momentum-based whole body control framework for floating base robots, the capturability-based high level control, and the application to the ‘Atlas’ humanoid robot. The core of this control framework is the combination of a quadratic program and a recursive Newton-Euler algorithm. From a set of desired motion constraints, rate of change of momentum objective, and ground contact information, the core calculates the desired joint torques to be applied on the robot. We’ll present how in the walking controller, these desired inputs are computed based on the capturability framework. Finally, the application to real hardware will be shown through the robot ‘Atlas’.
Twan Koolen, MIT
A QP control framework based on centroidal momentum and motion constraints
This presentation describes the QP-based whole-body control framework that was used in IHMC's Virtual Robotics Challenge entry as well as during the DRC trials, albeit in a modified form. The control framework allows desired robot behaviors to be specified as linear constraints on the joint acceleration vector, termed motion constraints. The QP finds a joint acceleration vector and external forces that reconcile the motion constraints with friction cone constraints, using a conservative polyhedral approximation. Exploiting the properties of whole-body centroidal momentum results in a small QP size. Strong links between instantaneous capture point dynamics, the Centroidal Moment Pivot, and the rate of change of centroidal momentum allow for simple high level motion planning.
Scott Kuindersma, MIT
An efficient controller for whole-body dynamic locomotion
We describe a dynamic walking controller implemented as a convex quadratic program (QP). The controller minimizes a control-Lyapunov function derived from a simple walking model while respecting the dynamic, input, and contact constraints of the full robot dynamics. Using a custom active-set algorithm to exploit the sparsity and temporal structure common to most QP-based walking controllers, we surpass the performance of the best available off-the-shelf solvers and achieve 1kHz control rates for a 34-DOF humanoid. We describe applications to balancing and walking with Atlas.
Alexander Herzog, MPI
Balancing experiments with a hierarchical inverse dynamics controller on a torque controlled humanoid
Recently several hierarchical inverse dynamics controllers based on cascades of quadratic programs have been proposed for application to torque controlled robots. They have important theoretical benefits but have never been implemented on a torque controlled robot where model inaccuracies and real-time computation requirements can be problematic. In this contribution we present an experimental evaluation of these algorithms in the context of balance control for a humanoid robot. The presented experiments demonstrate the applicability of the approach under real robot conditions (i.e. model uncertainty, estimation errors, etc). We propose a simplification of the optimization problem that allows us to decrease computation time enough to implement it in a fast torque control loop. We implement a momentum-based balance controller which shows robust performance in face of unknown disturbances, even when the robot is standing on only one foot. In a second experiment, a tracking task is evaluated to demonstrate the performance of the controller with more complicated hierarchies. Our results show that hierarchical inverse dynamics controllers can be used for feedback control of humanoid robots and that momentum-based balance control can be efficiently implemented on a real robot.
Luis Sentis and Gwen Johnson, UT Austin
Shared Control and Real-Time Performance Issues in Torque Controlled Humanoid Robots
This presentation addresses architecture design considerations for real-time performance which enabled the semi-autonomous control of the Valkyrie robot for the DARPA Robotics Challenge. A middleware with a basis in the Whole Body Control (WBC) task-prioritized controller--and later extended to support other algorithms--was designed for rapid re-parametrization of the controller by either on-line planning modules or human operators. To achieve the fast servo rates necessary for closed-loop control of high DoF systems, a multi-threading approach was used to optimize the software. We also discuss our experiences and practical considerations in closed-loop control of series elastic humanoid robots.