fortiss scientist Julian Bernhard receives the “Best Paper Award” for his research into interactive decision-making algorithms in autonomous vehicle systems.
The fortiss doctoral candidate received the award for his paper "Robust Stochastic Bayesian Games for Behavior Space Coverage". Co-author is Alois Knoll from the Technical University Munich (TUM). The publication was presented during the RSS2020 Interaction and Decision Making in Autonomous Driving workshop on July 13, 2020. The event, organized by the renowned American universities UC Berkley and Massachusetts Institute of Technology (MIT), was held as part of this year’s Robotics Science and Systems (RSS) conference.
The workshop, dedicated primarily to the topic of interactive decision-making in autonomous driving systems, was organized with the goal of providing a platform for discussing the most recent advances, current challenges and future direction of research in the field of robust decision-making for autonomous driving environments.
Given that increasing numbers of autonomous vehicles are operating on open streets, the development of robust decision-making algorithms is becoming increasingly important. With this in mind, the latest research efforts are focused on detecting, modeling and forecasting the behavior of traffic participants.
Julian Bernhard and his colleagues in the Trustworthy Autonomous Systems field of competence are developing interactive decision-making algorithms for autonomous vehicles that guarantee a dependable driving strategy even when the vehicle is confronted with previously-unknown human behavior. In order to be able to examine unknowns in advance of driving tests, fortiss is developing a simulation framework that it calls BARK, which will be introduced in October of this year during the International Conference on Intelligent Robots and Systems (IROS).
