AUTONOMOUS DRIVING SYSTEMS : CONTINUITY VS DISCRETE
The problem of autonomous Driving systems can be explained this way : engineers apply a “scenario-based method”. They cross characteristics of Driving situations and find an appropriate answer to every crossing. This work is very complex, because it involves “dynamics and time”. And if you try it for a demo, it works …
The question is “how can you validate that it actually works all the time ?”.
Yes if you want to let human driver quit the situation monitoring task to do something else, it seems to be a key question ? With a DISCRETE method such as the scenario approach it is obvious that you simply cannot answer this question. This is why NEXYAD proposed to use their real time driving risk computing API SafetyNex to build an ADAPTIVE Autonomous Driving System as an added layer to existing system : in the “holes” between crossings , the risk is still assessed (20 times per second) and you can then apply a continuous method : i.e. “when risk rises too much then slightly slow down and check is risk decreases. NEXYAD is currently helping valuable teams to do it and GO TO LEVEL 5.
We may do that quicker than most people think.
