University of Michigan (UMichigan) researchers are teaching self-driving cars to recognize and predict pedestrian movements with greater precision than conventional systems.
The system collects data from vehicles through cameras, LiDAR, and global-positioning system, allowing the researchers to capture video clips of humans in motion and then recreate them in three-dimensional computer simulations.
The researchers use those models to create a "biomechanically inspired recurrent neural network" that catalogs human gait, body symmetry, and foot placement.
The UMichigan system can predict poses and future locations for one or several pedestrians up to about 50 yards from the vehicle.
Said UMichigan's Matthew Johnson-Roberson, "We're training the system to recognize motion and making predictions of not just one single thing—whether it's a stop sign or not—but where that pedestrian's body will be at the next step and the next and the next."
From University of Michigan
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found