Luis Sentis

We are pleased to announce ControlIt!, a new open-source software framework for Whole-Body Operational Space Control. It is offered with an LGPL v2.1 open source license. Installation and usage instructions are available here:
ControlIt! Website.The draft paper describing the software framework can be found clicking below:

On the photo below: Liang Fok, Gwen Johnson, Luis Sentis, Nick Paine

The stabilizing properties of phase space planning in combination with the compliant SEA-based robot are shown for a N step task. Improvements on low level controllers allow the system to achieve its highest performance to date. Next improvement will be to enhance pose estimation.

The Valkyrie robot at IHMC taking some steps and doing one legged balance routine. The robot is designed and built by NASA JSC in collaboration with the University of Texas at Austin. Control algorithms by IHMC. Funding for IHMC and UT Austin provided through the NSF/NASA National Robotics Initiative.

In this video we show improvements on phase-space dynamic walking based on using an absolute return frame and the addition of Coriolis/centrifugal effects. The description of the planner can be found on Arxiv Preprint

Arxiv Preprint. In the video below, we develop estimation and control methods for quickly reacting to collisions between omnidirectional mobile platforms and their environment. To enable the full-body detection of external forces, we use torque sensors located in the robot’s drivetrain. Using model based techniques we estimate the location, direction, and magnitude of collision forces, and we develop an admittance controller that achieves a low effective mass in reaction to them. For experimental testing, we use a facility containing a calibrated collision dummy and our holonomic mobile platform.

Arxiv Preprint. The first experiment, shows the Hume biped robot balancing on a high pitch split terrain with and without push disturbances. We implement a Whole-Body Operational Space Controller to compute joint torques consistent with a desired set of operational space accelerations, known contact constraints, and desired internal forces. The internal forces, during multi- contact, correspond to the linear subspace of joint torques that do not cause accelerations of the robot. For undirected walking, Hume continuously steps forward and backward to remain balance. To accomplish this capability, we feed foot trajectories from an algorithm called Continuous Time to Velocity Reversal Online Planner. The planner continuously calculates new trajectories for the feet in an online fashion to recover from disturbances.