Thesis Seminar: Interaction Between Autonomous Vehicles and Pedestrians, 30th November, 1pm
When: Wednesday 30th of November, 1pm AEDT
Where: This seminar will be presented online via Zoom, RSVP here.
Speaker: Kunming Li
Title: Interaction Between Autonomous Vehicles and Pedestrians
Abstract:
Autonomous vehicles have been developed rapidly in recent years as they are believed to improve transportation efficiency and prevent accidents. Safe and efficient autonomous vehicles are expected to bring a range of potential benefits to society, such as reducing risky and dangerous human driving behaviours, avoiding crashes and improving productivity. While many autonomous vehicles have been tested on public roads, pedestrians’ interactions and behaviour are not comprehensively understood or thoroughly integrated for the safe and efficient navigation of autonomous vehicles.
Autonomously driving in a crowded pedestrian environment is still a major challenge. A reliable pedestrian perception system is the prerequisite for crowd navigation of autonomous vehicles, which requires autonomous vehicles to detect and track pedestrians accurately using limited on-board sensors. Furthermore, it is essential to understand the detected pedestrians’ behaviour and predict their intention in order to make safe driving decisions. This problem becomes more complex when we consider uncertainty, the multimodal nature of human motion as well as the implicit interactions between members of a crowd, including any response to a vehicle. Additionally, pedestrians often are not fully observed due to the limitations of on-board sensors, such as visual occlusion, resulting in inaccurate prediction and risky driving decisions.
This presentation focuses on how to enable autonomous vehicles to seamlessly and safely interact in close proximity to pedestrians by predicting the uncertain branching and multi-modal nature of pedestrian motion during social interactions. In addition, the challenge faced in partial observation environments is addressed in our work, where obstructions in the urban environment place limits on their visibility.
Bio:
Kunming Li is a final year Ph.D student in the Australian Centre for Field Robotics and the University of Sydney. Prior to joining ACFR, Kunming got a B.Eng. from the Australian National University. His research focuses on interaction between pedestrians and autonomous system, including modeling uncertainty of pedestrian motion, social navigation in crowded environments, and the pedestrians perception system for autonomous driving systems.