Smart Vehicle Technologies
Sensor fusion to Moovee means time-synchronized sensing, with all (surround) sensor outputs stitched onto a unified 4D (polar and time) format for next stage navigational processing. Geometric sensing data is collected in high resolution single pass, and then successively "filmed" in realtime as the vehicle motions. A static example is shown here. It must be highlighted that the resolution of this capture will be improved greatly as we advance in fitting side sensors differently than frontal and rear placements, together with corner sensings under road safety considerations.
Surround Sensing System
Vehicle Kinematic Control
Driverless vehicle requires x-by-wire mechatronic controls over motoring, braking, and steering. Full vehicular autonomy means machine operating without humanistic anxiety, instinct, or mannerism, and may overdrive existing vehicle designs into instabilities since many chassis are derived from overutilization past. Moovee has structured dynamic kinematic matrix for realtime controls and accounted for deteriorations in between maintenance schedules. We showcase here a simulation study of kinematic controls done in 2015.
Machine Learning Navigation
Artificial intelligence could be thought as the higher order complex mathematical/statistical modeling more superior than polynomial approximation in classical engineering formulations. It has an added advantage of self improvement when shaped with variance of scenarios. Moovee has applied ai techniques in at least four areas with vehicle navigation in real traffic being the most prominent. We present here a video demonstrating our learning algorithm (top LEFT corner) in comparison with actual drive-footage and steering operations.
For novice, the above video seems unextraordinary, but when comparing with field trial readied vehicular lidar system this static Moovee surround sensing version is already far exceeded expectations.