Communications of the ACM
Hand-Tracking Tech Watches Riders in Self-Driving Cars to See If They're Ready to Take the Wheel
University of California, San Diego researchers developed a hand-tracking system to determine how long it would take a driver to assume control of a self-driving car in an emergency.
Credit: Mikhail Pochuyev/Getty Images
Researchers at the University of California, San Diego developed a hand-tracking system to determine how long it would take a driver to assume control of a self-driving car in an emergency.
The researchers took an existing program for tracking full-body movements and adapted it to focus on the wrists and elbows of the vehicle’s driver and also a front-seat passenger, if present.
The team developed and applied machine learning algorithms to train the system to support Level 3 autonomous technology, then trained it with 8,500 annotated images.
They found the system was able to identify the location of each of eight elbow and wrist joints of both the driver and passenger with 95% accuracy, although it had a localization error of 10% when estimating the average length of driver or passenger arms.
From IEEE Spectrum
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found