The National Science Foundation’s Cyber-Physical Systems (CPS) program has awarded $1.2 million to a team led by Ricardo Sanfelice, professor of electrical and computer engineering at UC Santa Cruz, to develop and demonstrate algorithms capable of autonomously adapting their motion plan and control actions.
The project, “Constraint Aware Planning and Control for CPS systems,” is a collaboration between UC Santa Cruz and the University of Michigan. Sanfelice and his University of Michigan counterpart Professor Shai Revzen are the principal investigators. Funds will be divided equally between institutions, and distributed over three years.
The goal is to use aerial vehicles and walking robots to demonstrate a new fundamental science for computer controlled complex physical systems being developed by Sanfelice and Revzen. This new science will enable autonomous planning and control in the presence of failures and abrupt changes in system variables, such as rapid maneuvers in an emergency. Specifically the investigators will design a framework for algorithms that exploit awareness of physical and design constraints to autonomously self-adapt their motion plan and control actions.
“For algorithms to perform in real-world settings, the constraints imposed by physics and design specifications must be dealt with properly,” Sanfelice said. “The traditional paradigm for control design consists of separate layers for specification, planning and regulation but these algorithms struggle to adapt to anomalies when replanning and reconfiguration are required. In this project we will generate a framework for the co-design of planning and control algorithms that are capable of learning, adapting and reacting to changes.”
Their approach uses elements from geometry, adaptive control, and hybrid control to advance the knowledge on modeling, planning, and design of CPS with constraints, nonsmooth, and intertwined continuous and discrete dynamics. Unlike current approaches, which separate the task associated with planning the motion from the design of the algorithm used for control, the algorithms to emerge from this project will self-learn and self-adapt in real time to cope with unexpected changes in motion and specification constraints. This could help enable autonomous systems to perform robustly and safely, and degrade gracefully under failure conditions.
“In addition to new theoretical developments,” Sanfelice said, “through partnerships with the University of Bologna, Samsung, and Ghost Robotics, we will implement and validate our algorithms in real-world ground and aerial autonomous vehicles.”
For more information about Ricardo Sanfelice and his Hybrid Systems Laboratory, please visit: https://hybrid.soe.ucsc.edu/home
