Michigan University researchers develop technology to help self-driving cars recognize pedestrian movement more precisely

By utilizing video clips that runs for several seconds, researchers claim their system can study the first half of the snippet to make its predictions, and then verify the accuracy with the second half

Researchers at the University of Michigan (U-M) in the United States are teaching self-driving cars to recognize pedestrian movement more precisely by zeroing in on humans’ gait, body symmetry and foot placement, the university said in a press release on 12 February. These researchers are leveraging data collected by vehicles through cameras, lidar and GPS to capture video snippets of human in motion and then recreate them in three-dimensional (3D) computer simulation. Using that, the researchers have created what they call a “biomechanically inspired recurrent neural network” that catalogs human movements. With this, they can predict poses and future locations for one or several pedestrians up to about 50 yards from the vehicle, the researchers claimed.