A Cognitive Information Processing Pipeline for Multiple–Access Vehicular Camera Communication

The ability to perceive safety-critical events by virtually seeing through vehicles and other obstructions on the road can be a very useful driving assistance feature for vehicles. In this paper, we first hypothesize that such a feature can be achieved by vehicles driving ahead proactively communicate information about the visual scenery they perceive using dashboard cameras. Using such a multiple access setting, and considering brake-light light emitting diode (LED) to camera communication as the enabler, in this paper, we position an information processing pipeline for predicting the non-line-of-sight (NLOS) safety event. In particular, we present the algorithmic design of the cognitive information processing system that will continuously warn the host driver about the line-of-sight (LOS) and non-line-sight road situations. We position the use-case of our proposed information processing pipeline through a case-study analysis for a real-world driving scenario.