I couldn’t agree more with what Alan Winfield says about this problem: The challenge is not to have an autonomous car, the challenge is to prove the correctness of the system, and of a system that learns during the process. This type of adaptive systems pose several challenges and probably the most challenging one is the public perception of their quality and robustness. We’ve seen some problems recently: The cases of the electronic accelerator of the Prius or the emergency-breaking system of the Volvo C60, both show that these systems are in their early deployment days. To go from this stage to a future driving experience where our cars are our personal chauffeurs there’s still a long road ahead (pun intended), and probably we wont see these kinds of cars in the next 10 years or so.
Links: