This paper presents a novel transfer learning algorithm to achieve high-performance trajectory tracking with Unmanned Aerial Vehicles (UAVs). The authors exploit an existing Iterative Learning Control (ILC) algorithm based on the notion that the performance of a system that executes the same task multiple times can be improved by learning from previous executions. However, the learning phase needed to apply such technique is related to each specific system, thus making the application of ILC poorly scalable. To overcome this limitation, the authors propose an H∞-optimisation-based definition of a dynamical transfer map that allows transforming the input signal learnt on a source system to the input signal needed for a target system to execute the same task. A Monte Carlo analysis has been carried out with the aim of showing the performance improvements due to the transfer knowledge. Finally, the proposed approach has been validated through experimental activities involving two different-scale quadrotors performing an aggressive manoeuvre.
H∞-based Transfer Learning for UAV Trajectory Tracking
2022-06-21
3757666 byte
Conference paper
Electronic Resource
English
Optimization-based iterative learning for precise quadrocopter trajectory tracking
British Library Online Contents | 2012
|Vehicle trajectory tracking method and vehicle trajectory tracking system
European Patent Office | 2020
|Joint trajectory tracking and recognition based on bi-directional nonlinear learning
British Library Online Contents | 2009
|Virtual Evaluation of Deep Learning Techniques for Vision-Based Trajectory Tracking
SAE Technical Papers | 2022
|