Nonlinear Dynamics and Distributed Control for Soft Robot Locomotion Research Topic: 1.0 Mechanical Sciences: 1.3 Complex Dynamics and Systems
Abstract
The long-term goal of the proposed work is to develop models and associated control schemes for a soft robotic system that is designed to locomote on a range of surfaces. In order to advance understanding of the robot s dynamics, the design and fabrication of the robot will be intimately linked to the modeling and control efforts. This synthesis is motivated by the complementary needs of modifying the robot design so as to improve the fidelity of the models and minimizing the sensor requirements needed for a control scheme. The latter scheme features networked sensors and control that achieves synchronization and other global patterns using local connections. The aforementioned control scheme has the potential to advance the state of the art in soft robotic locomotion. However, before it can achieve its potential, a complementary set of questions needs to be addressed and the specific scientific issues the proposal addresses include the following: (i) What frictional laws and reduced-dimensional theories in mechanics (e.g., elastic rod theory, shell theory, plane-strain linear elasticity) best capture the mechanics of soft robots limbs, typical contacting substrates, and their tribo-elastic interactions? (ii) What are the natural boundary conditions that govern unilateral contact, and how are these influenced by the adhesion and dissipation arising from inelastic deformation or stick-slip behavior? (iii) As in natural organisms, how can friction and adhesion be actively tuned to control the tribo-elastic interactions of soft-rigid or soft-soft contacts and support dynamic locomotion or robust grasping? (iv) In the area of dynamics and control of a soft robotic system with distributed sensing and actuation, to what extent can the natural dynamics of a soft, flexible structure be exploited in its closed-loop operation? And (v) how might one apply tools from the control of networked systems to design a feedback control system for a soft machine with many spatially distributed sensors and actuators? The team at UCB will construct a hierarchical modeling framework to describe legged locomotion using nonlinear dynamics and continuum mechanics in a manner conducive to control. A hierarchy of three mathematical models will be developed featuring thermo-mechanical rod theories with a parallel effort on computer animations of the complex motions predicted by each model. Complementing this work, the team at CMU will perform an empirical and theoretical study of tribological interactions for ground contact with a soft robot limb. Because of their low mechanical rigidity, soft-bodied organisms and robots undergo significant deformation and conform to surfaces with which they make contact. This unilateral contact occurs over large areas of the body surface and introduces tribological effects that can strongly influence motion, power consumption, and internal stress distributions. Finally, the team at UMD is the design and implementation of a distributed nonlinear controller for legged locomotion using sensor feedback. This task will utilize the dynamic models from Task 1 and the soft-robot testbed constructed in Task 2. The focus is on a distributed estimation and networked-control framework with a coupled oscillator model for coordinated leg motion and hybrid spine control of intrinsic curvature, curvature, and geometric torsion. The project will develop high-fidelity, dynamic models for nonlinear control of a highly deformable body. These models will be far more sophisticated than those currently used in soft robotics. Indeed, since undulatory motion has many more degrees of freedom than traditional locomotion in robotic systems such as the robotic cockroach, a hierarchy of state-space models with clearly defined force inputs is proposed for analysis and control.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 01, 2019
- Source ID
- W911NF1610242
Entities
People
- Oliver O Reilly
Organizations
- Army Contracting Command
- United States Army
- University of California, Berkeley