A sensor system for Obstacle Detection, Classification and Tracking (ODCT) has been designed to help make autonomous rail transport more efficient, less costly to operate, and capable of 24/7 functionality while minimising accidents.
The OnTRAC project was a 30-month program to prototype and validate the feasibility and functionality of a state-of-the-art sensor fusion system, built upon proven LIDAR technology.
The system was developed by Lumibird Canada, a designer and manufacturer of LIDAR systems; global digital technologies leader Thales; and the Lassonde School of Engineering at York University, a Toronto-based research team with extensive expertise in 3D modelling of railway infrastructure.
It integrates with autonomous rail vehicles for the purpose of OCDT in varying weather conditions.
Beginning in 2019, the partnership first performed an investigative study of the challenges and threats posed to autonomous rail vehicles, then developed a novel sensor architecture, resulting in a new prototype LIDAR system specifically designed for fleet (rail) vehicles, and finally concluded the project in 2021 with safety and operational assessment – by way of in-situ rail demonstrations – in typical and adverse weather conditions.
Leveraging the field-proven OPAL 3D LIDAR product family, the successful completion of this project resulted in an integrated suite of different vision sensors (especially a novel, scalable, 3D LiDAR design), with innovative deep learning and artificial intelligence (AI) algorithms for object detection, classification and tracking in a systems architecture that addresses the stringent safety needs and performance requirements for autonomous rail operation.
The novel system established in this project will support the development of safe, autonomous urban rail systems, to make rail transport more efficient, less costly to operate, and capable of continuous functionality while reducing accidents due to lack of experience/training, fatigue, and other related effects which affect human operators.