Despite the great success in recent years in the area of perception and control of autonomous vehicles, most of the developed algorithms to date are limited to vehicles operating (e.g., driving or flying) at low-to-moderate speeds. Agility and maneuverability requires sensing and execution at much shorter time scales. Given these short response time scales, perception and action are becoming challenging tasks using existing route optimizers. This is especially true for small and agile UGVs and UAVs, which may not have the on-board computational capabilities (CPU and memory) to implement some of the sophisticated perception and path planning algorithms proposed in the literature thus far. Furthermore, during extreme maneuvering, the vehicle motion is coupled with data acquisition and sensing. New algorithms and methodologies are needed to tackle this problem and these methodologies most likely will span diverse areas beyond control theory: machine learning, artificial intelligence, real-time algorithms, information theory, compressive sensing, etc. Certifying and validating such algorithms is also a major challenge.
The objective of this workshop is twofold: the first objective is to report on current advances in the area of perception and control to enable “aggressive agility” for autonomous agents; the second objective is to bring together - in the same room - researchers from a diverse set of disciplines that are interested in this topic. Some of the same questions and problems posed above have very likely already been encountered (perhaps disguised) in other engineering fields. It is therefore imperative to start a more direct exchange of ideas and available methodologies between different researchers working in the neighboring fields of computer vision, machine learning, control, identification, communication all of whom may work on largely equivalent problems.
Potential participants to this workshop are control researchers who work in sensing, planning, or real-time embedded implementations and nonlinear control synthesis; computer vision researchers interested in information-theoretic, active perception and attention mechanisms; robotics researchers interested in nonlinear dynamics, stochastic optimal control and closure of perception/action loops in real-time for agile agents.