Matej Novosad presents Achieving Minimum-Time Flight: Real-Time Multi-Waypoint Planning for Agile UAVs

How can we push the limits of UAV speed and agility while ensuring real-time,
onboard computation? Join us in a short seminar as we unveil a breakthrough
approach to minimum-time trajectory generation for multirotor drones—one that
finally makes high-speed, multi-waypoint replanning feasible, even on limited
flight hardware.
Our method leverages a novel iterative thrust decomposition algorithm that
fully
utilizes a UAV's collective thrust - something previous point-mass models
failed
to achieve. By integrating gravity and drag modeling, we dramatically reduce
tracking errors in high-speed maneuvers. Paired with an efficient
gradient-based
optimization strategy, our approach computes minimum-time trajectories in
milliseconds while enabling accelerations of up to 3.5 g and speeds over 100
km/h.
We will showcase both simulation and real-world results, demonstrating that our
trajectories, executed via Nonlinear Model Predictive Control, match or even
outperform those generated in hours using full multirotor models. Our
open-source framework opens new possibilities for autonomous drone racing,
search-and-rescue, and next-gen aerial robotics.

Reference:
K. Teissing, M. Novosad, R. Penicka and M. Saska, "Real-Time Planning of
Minimum-Time Trajectories for Agile UAV Flight," in IEEE Robotics and Automation
Letters, vol. 9, no. 11, pp. 10351-10358, Nov. 2024, doi:
10.1109/LRA.2024.3471388
Paper: https://doi.org/10.1109/LRA.2024.3471388
Video: https://youtu.be/wArd536Amro
Code: https://github.com/ctu-mrs/pmm_uav_planner