Autonomy for Aerospace: Challenges, Solutions, and Research Directions 11 January 2021 1300 - 1415
Traditional autonomy stacks have not been equipped for the real complexities introduced by advanced perception systems on robotic platforms, such as operation utilizing limited range, noisy sensors in dynamic environments. This talk will overview existing technology, outstanding challenges, required capabilities, and recently developed solutions that integrate planning and perception to provide robust decision-making for real-world autonomy. The talk will also motivate and outline the development of safety and performance guarantees for learning-based approaches, along with methods for incorporating next-generation scene understanding capabilities into planners - all while ensuring that realistic uncertainties are appropriately captured in models used by autonomous systems.
Jonathan P. How
Richard C. Maclaurin Professor of Aeronautics and Astronautics, Massachusetts Institute of Technology