Zhuyi: Perception Processing Rate Estimation for Safety of Autonomous Vehicles
TimeTuesday, July 12th3:54pm - 4:18pm PDT
Location3000, Level 3
Event Type
Research Manuscript
Autonomous Systems (Automotive, Robotics, Drones)
Autonomous Systems
DescriptionThe processing requirement of Autonomous vehicles (AVs) for high-accuracy perception in complex scenarios can exceed the resources offered by the on-board computing platform, degrading safety and comfort. This paper proposes a sensor processing rate estimation model, Zhuyi, that quantifies the lowest safe processing rate in a driving scenario. Zhuyi can be employed post-deployment as an online safety check and to prioritize work. Experiments conducted using a multi-camera state-of-the-art industry AV system show that Zhuyi can reduce compute and bandwidth pressure by at least 2x compared to processing the camera feed at 30 Frames Per Second (FPS) in the tested scenarios.