Autonomous Vehicles

Research Overview

Autonomous vehicles are a compelling next-generation artificial intelligence technology. In order to safely navigate through the world, vehicles must plan long-range routes, make short-range plans, and perceive the world around them, all while acting under a policy that takes into account the likely future behaviors of agents in the surrounding environment. While not strictly AI-complete, the challenge of autonomous driving in urban and unstructured environments is substantial, as-yet unsolved, and of paramount economic importance.

The PLAI group is actively working with Inverted AI to develop new technologies based on advanced probabilistic machine learning and multi-agent behavior modeling. This collaboration specifically focuses on improvements to generative behavioral modeling with an application focus on autonomous driving; while also recognizing additional applications in assistive and advanced robotic applications and video game non- playable character development. Methodologically, we seek robust and computationally efficient solutions to learning and inference problems related to modeling complex multi-agent behavior over challenging time horizons. The concrete autonomous driving aim of this research is to build deep generative models that model what various agents near and on the roadway will do far enough into the future to allow for planning algorithms to ensure that autonomous vehicles can behave safely.

Contributors

PI: Frank Wood (UBC, CEO Inverted AI)

Lead Investigators (Inverted AI): Adam Ścibior (CTO), Berend Zwartsenberg (CSO)

Student Internships: Dylan Green, Jonathan Wilder Lavington, Justice Sefas, Ke Zhang, Matthew Niedoba, Saeid Naderiparizi, Vasileios Lioutas, Xiaoxuan Liang, & Yungpeng (Larry) Liu

Supporters

MITACS

Publications