Loading Events

ESE Fall Colloquium: “Robust Multi-Agent Autonomy Under Uncertainty”

October 15 at 12:00 PM - 1:00 PM
Details
Date: October 15, 2026
Time: 12:00 PM - 1:00 PM
Event Category: ColloquiumSeminar
Event Tags:
  • Tags:, ,
  • Organizer
    Electrical and Systems Engineering
    215-898-6823
    eseevents@seas.upenn.edu
    View Website
    Venue
    Raisler Lounge (Room 225), Towne Building 220 South 33rd Street
    Philadelphia
    PA 19104
    Google Map

    Unmanned ground and aerial systems hold promise for critical applications, including search and rescue, environmental monitoring, and autonomous delivery. Real-world deployment in safety-critical settings, however, remains challenging due to GPS-denied operation, perceptual uncertainty, energy constraints, and the need for safe trajectory planning in dynamic unknown environments. This talk presents recent advances in planning, control, and perception that together enable robust, scalable, and efficient aerial autonomy. On the planning and control side, I introduce SANDO, which enables uncertainty-aware trajectory planning for safe, real-time flight in dynamic and unknown environments, and MIGHTY, which performs fully coupled spatiotemporal optimization for agile and precise motion. I also present Dispersive Forward Tree search (DFT*), a near-optimal sampling-based planning framework whose massively parallel implementation enables real-time planning with high-fidelity vehicle models. Its multi-robot extension, Dispersive Tree Agreements, enables distributed coordination robust to imperfect communication. I then consider how aerial platforms can increase range and endurance by extracting energy from the atmosphere. I present ASAP, a soaring planner that exploits orographic and thermal lift while adapting to changing atmospheric conditions, along with robust and learning-based approaches for thermal and dynamic soaring. On the perception side, I present complementary mapping frameworks for long-term autonomy and planning. ROMAN builds sparse, open-set object maps that enable accurate relocalization under extreme viewpoint changes, while Meridian combines object and line features to localize ground robots in aerial imagery. GRAND SLAM combines 3D Gaussian splatting with semantic and geometric priors to produce unified scene representations suitable for photorealistic planning. I also discuss the interaction between perception and control, focusing on safety filtering for systems that rely on learned perception models. Finally, I present results from simulation and hardware experiments and conclude with open challenges in building resilient multi-robot autonomous systems.

    Speaker

    Jonathan How

    Jonathan P. How

    Ford Professor of Engineering, MIT

    Jonathan P. How is the Ford Professor of Engineering at the Massachusetts Institute of Technology. He received a B.A.Sc. in Aerospace Engineering from the University of Toronto in 1987 and S.M. and Ph.D. degrees in Aeronautics and Astronautics from MIT in 1990 and 1993. After serving as a postdoctoral associate at MIT and an assistant professor at Stanford University, he joined the MIT AeroAstro faculty in 2000. Professor How is President-elect of the IEEE Control Systems Society. He was elected to the National Academy of Engineering in 2021 and is a Fellow of IEEE and AIAA.