The People’s Autonomous Vehicle
Visual perception: a combination of semantic information extraction through deep learning technologies and spatial depth extraction through traditional computer vision techniques.
Visual inertial odometry (VIO): we fuse the four camera sensors and the inertial measurement unit to generate accurate real-time position updates.
GPS/VIO fusion: GPS is great sensor for localization but at times we do not receive good GPS signal, leading to localization failures. By fusing GPS and VIO, we could achieve accurate and real-time localization results even in GPS-denied situations.