Autonomous Auto – Men and dogs

Reply to this post

I see a shake out with only a handful surviving. 

• Just as in 2011 when every man and his dog was building an app store, now everyone is trying to build an autonomous vehicle.
• Just as it was with app stores and every other technology craze before it, only a very few are likely to succeed.
• Google, Apple, Tesla, Baidu, Alibaba, Tencent, Uber, Lyft, Faraday Future, Didi Kuadi, Mobileye, LeEco as well as every company that already makes cars are all trying to come up with an autonomous car.
• Of all of these, I think Google is the most advanced as it has been working on self-driving cars the longest and I believe it has the best machine learning and artificial intelligence heritage.
• I have long been of the opinion that autonomous cars are going to take far longer than anyone expected to become mainstream (see here).
• Unfortunately, the internet industry has taken the example of the time that it took to become a global force and unrealistically applied the same logic to autonomous cars.
• The automobile industry is heavily regulated both in terms of its products and how those products are used by consumers.
• Hence, there is every sign that the technology will be ready long before the market is giving plenty of time for a shakeout before the real roll out begins (see here).
• In order for an autonomous solution to work a number of things need to come together.
  • First. A highly accurate map (although some argue that highly accurate sensing is enough).
  • The current maps are not accurate enough and so the mapping companies are going through the process of re-mapping the world in high definition.
  • This is an expensive process and I suspect that only Google, HERE and Apple (if it so desires) have the means necessary to do this.
  • China is likely to be re-mapped by Chinese companies and remain a law unto itself.
  • Second. A highly accurate positioning and sensing system for the car itself.
  • A high definition map is useless unless the car knows exactly where it is and where the obstacles are and the GPS system in use today is not nearly accurate enough.
  • Third. A sophisticated artificial intelligence system that is capable of dealing with unpredictable events that will be caused with non-autonomous vehicles sharing the road.
• Furthermore, all of these systems will need to be somewhat compatible with each other as at some point they will need to communicate with server based traffic management systems.
• This is because there are huge savings to be made in journey times if there is a centralised system that is aware of where all the vehicles are and where they are going and can manage the flow accordingly.
• I suspect that there will be a handful of autonomous driving systems that make it with the vast majority falling by the wayside.
• Those that don’t have it will have to licence it, but for the smaller players this will make more sense as they are unlikely to be able to afford the huge cost of developing an entire system themselves.
• This creates an opportunity for those that don’t make cars such as HERE, Mobileye and Google to come up with a neutral offering that can be licenced to anyone that wants it.
• The auto industry is rightfully wary about Google’s intentions (see here) when it comes to Android Auto but if it is more open with regards to the data it collects, there could even be the possibility for car companies to get its system for free in exchange for some data sharing.
• Either way, it is likely to be a very long time before autonomous driving achieves the kind of reality that will meaningfully affect people’s lives and the economy.
• I expect the shake out to begin in 2018 once those developing these systems realise that they will not be able to meet the promises they have made to their backers.