Automated synchronization of driving data using vibration and steering events

摘要: A passive synchronization method for driving data is proposedSynchronization of vehicle sensors uses vibration and steering events.Dense optical flow video used to capture significant car vibrations events.Cross correlation sensor pairs achieves 13.5ms accuracy. We propose a automated useful the study multi-modal driver behavior design advanced assistance systems. Multi-sensor decision fusion relies on synchronized streams in (1) offline supervised learning context (2) online prediction context. In practice, such are often out sync due absence real-time clock, use multiple recording devices, or improper thread scheduling buffer management. Cross-correlation accelerometer, telemetry, audio, dense from three achieve an average error 13 milliseconds. The insight underlying effectiveness proposed approach that described overlapping aspects allowing cross-correlation function serve as way compute delay shift each sensor. Furthermore, we show decrease duration stream.