PECASE (topic 1.3.2): The Case for Morphologically Modulated Dynamics
Abstract
The objective of the project is to provide predictive models of animal and robot locomotion. We are focused on capturing the robustness and survivability of the most reliable terrestrial locomotion systems knownland-based arthropodsand transferring this essence into new mathematical representations and instantiation techniques, resulting in robotic technologies that move in fundamentally more robust ways than existing legged systems (be they for transport of soldiers and materiel in the field, or for improved prosthetic devices back home). This work builds on fundamental advances in the theory of hybrid dynamical systems and coupled oscillators (informed by insights gained through a previous ARO YIP grant). The initial phases of this work consisted of publishing results from previous work and applying these findings to establish a new robot platform for further experimentation. We have continued to advance our methods for identifying the body dynamics of robots and animals while developing more advanced forms of instantiating and modulating dynamics. We also continued work on connection-based models. These methods form the foundation of a coherent methodology for producing robots whose body morphology contributes to simplifying their control, and whose software can identify and reinstantiate desired behaviors if they fail. Our approach is grounded in oscillator theory, and accounts for both the crucial role of body morphology and mechanics in the dynamics of motion, and the interactions of the body with the environment. This fundamental progress in anchoring templates is proving to be a novel approach to control, applicable anywhere the control of high-dimensional nonlinear systems is of interest to the Army.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 30, 2019
- Accession Number
- AD1094527
Entities
People
- Shai Revzen
Organizations
- University of Michigan