Adding Safety to Probabilistic Movement Primitive Control via Control Barrier Functions by Dr. Nicholas Gans : vTools Events

IEEE.org | IEEE Xplore Digital Library | IEEE Standards | IEEE Spectrum | More Sites

Adding Safety to Probabilistic Movement Primitive Control via Control Barrier Functions by Dr. Nicholas Gans

#control #methodologies #systems #trajectory #learning #robotics #automation #probabilistic #movement #primitive #ProMP #Lyapunov #Function #CLF #barrier #function #CBF #intelligent #autonomy #autonomous

Webinar: Adding Safety to Probabilistic Movement Primitive Control via Control Barrier Functions by Dr. Nicholas Gans, Division Head of Autonomy and Intelligent Systems, the University of Texas at Arlington Research Institute.

Organized by:

IEEE Vancouver Joint Control, Robotics & Cybernetics Chapter

https://vancouver.ieee.ca/cs-ra-smc/

Date: February 10th, 2022

Time: 5:00pm to 6:30pm

Everyone Welcome. Please Register, Thank you.

Abstract:

In this seminar, Dr. Nichilas Gans will present a recent investigation into novel control methodologies that employ probabilistic movement primitives (ProMPs), control Lyapunov Functions (CLFs) and control barrier functions (CBFs).

ProMPs are a powerful tool for defining a distribution of robot trajectories via demonstration. However, the native ProMP control methods suffer from a number of drawbacks. For example, these methods tend to rely on linear control designs thus limiting their applicability in robotics. In addition, they tend to be overly sensitive to initial parameters which can lead to stability issues. Conversely, CLF and CBF control approaches employ feedback linearization to handle nonlinearities in the system dynamics and real-time quadratic programming to find an optimal control that satisfies all safety constraints while minimizing control effort. However, CLFs and CBFs remain difficult to define and implement without expertise in nonlinear control. We propose to define CLFs to regulate a system described by the ProMP mean and CBFs as function of the ProMP standard deviation. Thus, the system, such as a robot, may move along a trajectory within the distribution while guaranteeing that the system state never leaves more than a desired distance from the distribution mean. We then extend this approach to include time-varying CBFs which can be incorporated to avoid static and moving obstacles and investigate model predictive control approaches to optimize and accommodate constraints over a finite horizon. Furthermore, we highlight how the proposed method may allow a designer to emphasize certain safety objectives that are more important than the others. A series of simulations and experiments demonstrate the efficacy of our approach and show it can run in real time.

Bio: Dr. Nicholas Gans is Division Head of Autonomy and Intelligent Systems, the University of Texas at Arlington Research Institute. Prior to this position, he was a professor in the department of Electrical and Computer Engineering at The University of Texas at Dallas. His research interests are in the fields of robotics, nonlinear and adaptive control, machine vision, and autonomous vehicles. Dr. Gans earned his BS in electrical engineering from Case Western Reserve University in 1999, then his M.S. in electrical and computer engineering in 2002 and his Ph.D. in systems and entrepreneurial engineering from the University of Illinois Urbana-Champaign in 2005. He was a postdoctoral researcher at the University of Florida and as a postdoctoral associate with the National Research Council and Air Force Research Laboratory.