Interpretable AI for Seamless Robot Navigation Among People

This project aims to make advancements in mobile robot navigation in crowded environments, which will alleviate one of the key bottlenecks to wider utilization of mobile robotic systems and bring these systems into sectors such as entertainment, travel, health care, and logistics. To achieve this aim, the project proposes to develop robust, interpretable AI for human navigational intent prediction that, given input from a stereo camera, will output the probability distribution over where a person may be positioned the near future. Such predictions can be integrated with trajectory planning algorithms to allow robots to navigate much more quickly and smoothly around people without sacrificing safety. These planning algorithms will also be able to account for human reactions to potential robot actions. The efficacy of the proposed robot navigation system will be demonstrated in a real-world robotic system navigating among crowds.

Faculty Supervisor:

Mo Chen

Student:

Partner:

Ma Robot

Discipline:

Computer science

Sector:

Information and cultural industries; Professional, scientific and technical services

University:

Simon Fraser University

Program: