Toward Collective Intelligence for Resource-Constrained Autonomous Vehicles

The practical deployment of autonomous, semi-autonomous and teleoperated vehicles in contested environments faces several technical challenges connected to the scarce communication and computing resources available to these systems. On the one hand, small vehicles – e.g., expendable, and attritable UxVs – have obvious hardware, energy, and memory constraints that limit their ability to execute complex mission logics. On the other hand, offloading the execution of these logics to compute-capable devices require the transmission of information-rich signals over capacity limited and volatile wireless channels. Similarly, larger vehicles face challenges in processing and fusing the large volume of data produced by the many onboard sensors needed to make them mission-capable in a broad range of conditions, as well as in transporting them to remote operators and control centers. In this talk, I will provide an overview of the techniques and frameworks that my research group developed to allow flexible, efficient, and resilient distributed neural computing for robotic perception and autonomous navigation. Our approaches deeply integrate system solutions and machine learning to obtain practical frameworks deployable on real-world hardware platforms and applications. Speaker(s): Prof. Levorato Agenda: Social Hour: 6:00 p.m. Dinner: 6:30 p.m. Presentation: 7:15 p.m. Lakeside Orange County Airport Hotel, 7 Hutton Centre Drive, Santa Ana, California, United States, 92707-5794