Online Intention Prediction via Control-Informed Learning
This addresses the challenge of real-time intention estimation for autonomous systems like drones, though it appears incremental as it builds on existing inverse control/RL methods with online adaptations.
The paper tackles the problem of predicting time-varying intentions of autonomous systems with unknown parameters by formulating it as an inverse optimal control/reinforcement learning task, achieving accurate and adaptive prediction in simulations and quadrotor experiments.
This paper presents an online intention prediction framework for estimating the goal state of autonomous systems in real time, even when intention is time-varying, and system dynamics or objectives include unknown parameters. The problem is formulated as an inverse optimal control / inverse reinforcement learning task, with the intention treated as a parameter in the objective. A shifting horizon strategy discounts outdated information, while online control-informed learning enables efficient gradient computation and online parameter updates. Simulations under varying noise levels and hardware experiments on a quadrotor drone demonstrate that the proposed approach achieves accurate, adaptive intention prediction in complex environments.