US-based Southwest Research Institute (SwRI) has developed a motion prediction system that enhances pedestrian detection for automated vehicles.
The computer vision tool uses a novel deep learning algorithm to predict motion by observing real-time bio-mechanical movements with the pelvic area being a key indicator for changes.
“For instance, if a pedestrian is walking west, the system can predict if that person will suddenly turn south,” said SwRI’s Samuel E. Slocum, a senior research analyst who led the project.
“As the push for automated vehicles accelerates, this research offers several important safety features to help protect pedestrians.”
Although previous technologies could track and predict movements in a straight line, they were unable to anticipate sudden changes.
Motion prediction often uses optical flow algorithms to predict direction and speed based on lateral motion.
Optical flow, a type of computer vision, pairs algorithms with cameras to track dynamic objects, however accuracy reduces when people move in unexpected directions.
SwRI compared optical flow to other deep learning methods, optimising a novel temporal convolutional network that outperformed competing algorithms, predicting sudden changes in motion within milliseconds.
The temporal design uses a convolutional neural network to process video data, with SwRI optimising dilation in network layers to learn and predict trends at a higher level.
The research team leveraged SwRI’s markerless motion capture system, which uses camera vision and perception algorithms to provide deep insights into kinematics and joint movement.
Applications for the project, titled Motion Prediction from Sparse Skeletal Features, include human performance, automated vehicles and manufacturing robotics.
