Autonomous path planning and navigation in social environments
Martin Gerdzhev, Ph.D. Candidate
The goal of the project is to be able to dynamically plan and navigate in complex social environments, such as in shopping malls and other environments with heavy flow of people. Currently, traditional path planning algorithms do not take into account the complexities of moving objects, which leads to jerky motion and or no motion around such objects. This will allow robots to be deployed in a variety of urban settings where they were not previously able to go.
People detection, tracking, and following in social environments
Angus Leigh, M.Sc. Candidate
I'm currently developing methods to improve the wheelchair's ability to autonomous navigate in crowds and around people. Right now, the wheelchair's computer sees people as regular, stationary obstacles, and has no ability to understand who they are and how they move. If we can create intelligent algorithms that improve the wheelchair's perception capability in this area, it would greatly increase the functionality of the chair. It would also allow us to implement other interesting features, such as autonomous person following.
Metrics for robot evaluation in social environments
Andrew Sutcliffe, M.Sc. Candidate
The goal of this project is to find a way to objectively and quantitatively evaluate robot navigation behaviour in crowded spaces without the need of a human expert. The motivation behind this work comes from the growing interested in developing robots to assist people with mobility impairments. Traditional metrics for performance are maladapted in this case as they do not capture any social dynamics necessary to proper behaviour of autonomous navigation in crowds.