The deployment of autonomous robots in real-world settings often requires planning and decision making under multiple competing objectives. Which trade-off between objectives is the most adequate often depends on the respective end-user. Thus, a central problem is how inexperienced users can adapt autonomous systems to their preferences. In this talk we focus on two parts of the problem. First, we study human-in-the-loop learning frameworks that allow users to customize robot behavior to their preferences through a sequence of simple interactions. Once such interaction mode is choice feedback where users choose between two presented options, allowing for effectively learning user preference within few iterations. In the second part, we study the challenge of exploring optimal trade-offs for multi-objective robot planning problems. We establish fundamental theoretical properties that allow for efficient algorithm design. Applications of the presented work include high-level motion planning in human-centered environments, manipulation in servicing tasks, environmental monitoring, vehicle routing, and multi-robot pickup and delivery.
Nils Wilde is currently a postdoctoral fellow in the Autonomous Multi-Robots Lab at TU Delft, working with Javier Alonso-Mora. Until 2021 he was a postdoctoral fellow at the Autonomous Systems Lab at the University of Waterloo where he also did his PhD in Electrical and Computer Engineering under the co-supervision of Dana Kulić and Stephen L. Smith from 2016 to 2020. Before that he completed his BSc. and MSc. degrees at the Technical University Berlin in 2012 and 2016, respectively. Nils' research combines robot motion planning, multi-robot coordination, and human-robot interaction (HRI), developing algorithmic frameworks on the intersection of control, learning and optimization.