[This seminar is co-sponsored by the IEEE-CSS]
The complexity of modern autonomous systems has grown
exponentially in the past decade. Today's end-to-end control
demonstration have reached an impressive level of performance with
apparently no modeling effort. Yet, delivering high performance
autonomy which is safe despite environment uncertainty seems to be
critically linked to the ability of modeling and predicting the
uncertain environment.
Our research over the past decade has focused on control design for
autonomous systems which systematically incorporate predictions and
learning while guaranteeing safety. In this talk I will provide an
overview of the theory and tools that we have developed in this area,
address comparison with model-free approaches, and discuss key open
questions.
More info on: www.mpc.berkeley.edu
…Read more
Less…
