Making self-driving cars is one of the great AI challenges of the 21st Century and it involves many different parts. The goal is not merely to make fully autonomous driving cars so that humans never need to drive cars again. In fact, there are many forms of automation to every aspect of driving and coordination of vehicles on the road that can be considered.
In my lab we have done work on a few focussed topics in this area:
- Multi-Vehicle Communication - In a coordinated, multi-vehicle scenario such as a convoy or fleet or autonomous cars, it is important for the autonomous cars to communicate efficiently and reliably. In this topic we have looked at some ways to do this using Deep Neural Networks.
- Driver Behaviour Learning - In this line of research we look at how humans drive and try to learn models of that which can be predictive with a good level of accuracy. If autonomous vehicles drive in ways similar to, although hopefully safer than, humans, then they can more easily be integrated into the existing roads and traffic.
Our Papers on Autonomous Driving
- patentMulti-Level Collaborative Control System With Dual Neural Network Planning For Autonomous Vehicle Control In A Noisy Environment2020
- Deep Multi Agent Reinforcement Learning for Autonomous DrivingIn Canadian Conference on Artificial Intelligence 2020
- Learning Multi-Agent Communication with Reinforcement LearningIn Conference on Reinforcement Learning and Decision Making (RLDM-19) 2019
- Training Cooperative Agents for Multi-Agent Reinforcement LearningIn Proc. of the 18th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019) 2019
- Integration of Roadside Camera Images and Weather Data for monitoring Winter Road Surface ConditionsIn Canadian Association of Road Safety Professionals (CARSP) Conference 2019
- Decision Assist for Self-Driving CarsIn Canadian Conference on Artificial Intelligence 2018