RSS WS : Challenges in Dynamic Legged Locomotion
Saturday, July 15, 2017
Safe and Efficient Dynamic Robotic Locomotion.
California Institute of Technology
Humans have the ability to locomote with deceptive ease, navigating everything from daily environments to uneven and uncertain terrain with efficiency and robustness. With the goal of achieving these capabilities on robotic systems, this talk will present a unified formal framework for realizing dynamic behaviors in an efficient, provably correct and safety-critical fashion, along with the application of these ideas experimentally on a wide variety of robotic systems. In particular, we will introduce an optimization based control framework that is able to dynamically balance control objectives and safety constraints for dynamic robotic systems. These concepts will be illustrated through their application to the humanoid robot DURUS, with the result being dynamic and efficient locomotion displaying the hallmarks of natural human walking: heel-toe behavior. The translation of these ideas to robotic assistive devices and specifically powered prostheses will be described in the context of custom-built hardware. Finally, the extension of these concepts to safety-critical systems—including automotive applications, multi-agent systems, and swarms of quadrotors—will be discussed.
The Adaptive Significance of Robotic Gait Selection.
University of Michigan
Should a legged robot use different gaits at different speeds? If so, what constitutes these gaits? Where are they originating from and how should a robot be designed to exhibit them? In my presentation, I will explore these and related questions and presenting work that ranges from conceptual models to actual hardware implementations.
Design and Control for Dynamic Locomotion.
University of California Santa Barbara
Trajectory optimization is an increasingly practical and popular approach to obtain dynamically feasible and energy efficient nominal motions for legged robots. In this talk, we discuss complementary issues of evaluating and optimizing robot parameterizations (e.g., inertial properties) and feedback control laws (e.g., stiffness), toward improving performance goals such a controllability and robustness.
Legged Robots for the Field.
This talk provides an insight into our recent work to create four-legged robots that can operate under harsh conditions. I will outline some design aspects and general concepts that were followed to create a platform, which is simple to use and maintain. This includes work on compact, precisely torque controllable and impact robust actuator modules as well as on an overall system architecture aiming at versatile applications of a legged transporter. I will present a number of control, environment perception, and motion planning tools for static and dynamic locomotion in non-flat terrain and discuss all results in the context of extensive experiments that were conducted under realistic conditions in the field.
MIT Cheetah: New Design Paradigm for Mobile Robots.
Massachusetts Institute of Technology
Recent technological advances in legged robots are opening up a new era of mobile robotics. In particular, legged robots have a great potential to help disaster situations or elderly care services. Whereas manufacturing robots are designed for maximum stiffness and precision, mobile robots have a different set of hardware/software design requirements including dynamic physical interactions with environments. Events such as the Fukushima power plant explosion highlight the need for robots that can traverse various terrains and perform dynamic physical tasks in unpredictable environments, where robots need to possess compliance that allows for impact mitigation as well as high force capability. The talk will discuss several issues in mobile robot design focusing on the actuator characteristics. As a successful embodiment of such paradigm, the talk will introduce the constituent technologies of the MIT Cheetah, capable of running up to 13mph with an efficiency rivaling animals and capable of jumping over an 18-inch-high obstacle autonomously. A new version of the MIT Cheetah robot designed for real application will be introduced.
Foundations of Legged Locomotion.
Oregon State University
Creating walking and running devices is a complex endeavor. In this talk, I will attempt to string together a train of logical steps that point a direction for creating legged machines, based on the specific perspective of legged locomotion as a “dynamical phenomenon” that we as a community are gradually discovering. The talk will begin with a list of characteristics for a walking and running system (i.e. compliant interaction with the world, discontinuous contact, efficient energy cycle), with justifications for each. I will continue with design constraints to create the desired characteristics; put another way, I will discuss which aspects of the behavior should be implemented with passive elements (such as springs) and which aspects are better done in software. The goal of this talk will be to engage in ongoing discussion.
Dynamic Mobile Manipulation.
Legged robots equipped with manipulators and the control systems to coordinate them open a world of opportunity for robotics. Such mobile manipulation systems have greater workspace and reach than stationary robots, even when the arm mechanism is simpler. In this talk, I will give a status report on work Boston Dynamics is doing in this important new area.
Approximate Explicit MPC for Multi-Contact Feedback Control.
Massachusetts Institute of Technology
For robots governed by smooth, controllable (possibly nonlinear) ODEs, linear optimal control provides an extremely powerful tool for local stabilization of a fixed point or trajectory. Convenient parameterizations (e.g. LQR) allow designers to easily tune in the desired performance on a wide variety of systems, and the algorithms for computing the feedback gains are scalable and reliable. The state of the art in multi-contact feedback control is much worse — we mostly rely on one-step lookahead (e.g. QP inverse dynamics) and or hand-designed state machines — the major missing piece is a systematic design procedure that can reason about stabilization “across contact modes”.
In this talk, I will argue that piecewise-affine approximations of the system dynamics in floating-base coordinates are the natural and reasonable analog to linearization when locally stabilizing fixed-points or trajectories that lie on or near contact-mode boundaries. For these models, we actually know quite a bit about the feedback controller that optimizes a quadratic regulator objective: for instance, the controller is piecewise affine and has a piecewise-quadratic cost-to-go function. The problem is primarily that the number of pieces quickly becomes too large for even moderately-sized problems. I will present a number of algorithms that we have been developing to reliably approximate this optimal policy, often leveraging the idea that sampling from the optimal policy can be done as an MIQP, and discuss the prospects of actually having a turnkey solution that is anywhere near as powerful as LQR for smooth systems.