Parakh Manoj Gupta presents CurviTrack: Curvilinear Trajectory Tracking for High-speed Chase of a USV

In this seminar, we present and explain how we tackled the problem of landing a
UAV on a moving USV as opposed to the traditional approach of landing on a
stationary USV. The talk sheds light on evaluations and shortcomings of
existing
physics models for the prediction of USV motion. We then discuss how our
horses-for-courses approach which, using simple water drag and curvilinear
motion physics, can outperform a more complex state-of-the-art model. This
seminar is about simplicity and the performance gain these principles can
unlock. Moreover, we show how this model converges for a camera-based visual
pose estimation and predicts better than the state-of-the-art model requiring
various sensors providing full odometry. Our simulation and real-world
experimental results show how this approach benefits our goal of landing on a
curving USV and improves prediction performance by as much as 40%.
The seminar will be held in a short format of 20 minutes + 10 minutes Q&A.

Reference:
Gupta, P. M., Procházka, O., Nascimento, T., & Saska, M. (2025). CurviTrack:
Curvilinear Trajectory Tracking for High-speed Chase of a USV. IEEE Robotics
and
Automation Letters, (99), 1-8.
Paper: https://doi.org/10.1109/LRA.2025.3546079
Video: https://youtu.be/hGzNyRGAZSk