The Transportation Forecasting Competition (TRANSFOR22) was held in January this year and encouraged the participants to develop an algorithm that predicts pedestrian crossing times at intersections.
To be precise, the main goal was to protect vulnerable road users like people in wheelchairs or elderly pedestrians by extending the signal time, giving them more time to safely cross the street. The participants were provided with a signal performance log from a traffic controller and a Lidar dataset. The Transportation Research Institute of the University of Michigan won the competition by calculating the trajectory of a pedestrian, based on previous sample data. The Smart Transportation Applications and Research Lab of the University of Washington won the second place with a different approach. They predicted average crossing speeds of pedestrians and identified "higher-risk" pedestrians just from a three-second detection window. The AI x Mobility Lab of the Korea Advanced Institute of Science and Technology reached the third place using yet another method. They created an arrival-time prediction model based on the classification of pedestrians on foot and pedestrians in a wheelchair. With this model, they achieved 93% classification accuracy.
Ouster and four other sponsors made this competition possible by supporting the event financially. Therefore they are proud to be part of TRANSFOR22 and the future of enhancing road safety.
For more information, read the original article with detailed graphics and a closer description published by Ouster: https://ouster.com/blog/using-ouster-lidar-data-to-advance-intersection-safety-research/
