I’m at the Connected and Charged Symposium in Palo Alto today and there are a lot of very smart people all discussing the future of transportation. The problem is that many people here are talking about self driving cars in the wrong way. They’re thinking about them as though they are a single monolithic piece of technology and they aren’t. Self-Driving cars aren’t even a specific collection of scores of technologies working in concert. Rather, self-driving cars are a product feature set, which can be generated by a multitude of different approaches and architectures.
Someone asked John Hsu from Hyundai Ventures if autonomous vehicle development could be sped up if all of the companies working on AV platforms were sharing their testing data. The answer is NO for two reasons: (1) The architecture for Ford’s system, Tesla’s system, Apple’s system, etc are all different and so comparing the data would to some degree comparing apples and oranges and (2) Those systems should be different and should be developed independently. There is a ton of social/technological value in having those systems developed independently. We (the consumer) will benefit from seeing how well each of the AV approaches deals with the myriad of challenges presented by the real world. Some will invariably do better than others in rough weather, some will work better in heavy commute situations, some will work better in areas with old infrastructure and others will deal expertly with pedestrians. The process of technological evolution / hybrid vigor will eventually deliver the best system to the public.
Instead of thinking of Self Driving Cars as a single thing I think its beneficial to think about all of the underlying pieces of technology. Think about the sensor systems, the machine learning software/control systems, the onboard processing systems, the mapping systems, the robotic systems, the UI systems, the cyber security systems, the interconnectivity systems, etc. There are so many components that go into making autonomous vehicles work and we need to keep an open mind about how best to combine those components if we want to end up with powerful and flexible self-driving cars in the short to medium term.