RAMS Analysis of a Bio-inspired Traffic Data Sensor ("Smart Eye")

The Austrian Research Centers have developed a compact low-power embedded vision system "Smart Eye TDS", capable of detecting, counting and measuring the velocity of passing vehicles simultaneously on up to four lanes of a motorway. The system is based on an entirely new bio-inspired wide dynamic ¿silicon retina¿ optical sensor. Each of the 128 × 128 pixels operates autonomously and delivers asynchronous events representing relative changes in illumination with low latency, high temporal resolution and independence of scene illumination. The resulting data rate is significantly lower and reaction significantly faster than for conventional vision systems. In ADOSE, an FP7 project started 2008 (see acknowledgment at the end of the paper), the sensor will be tested on-board for pre-crash warning and pedestrian protection systems. For safety-related control applications, it is evident that dependability issues are important. Therefore a RAMS analysis was performed with the goal of improving the quality of this new traffic data sensor technology, in particular with respect to reliability and availability. This paper describes the methods used and the results found by applying a RAMS analysis to this specific case of a vision system.