Autonomous Autos – Glacial start.

Waymo does not want to be beaten by Yandex.

• Waymo is removing drivers in its commercial autonomous fleet just in time to be on par with Yandex which is about to do the same at two sites in Russia.
• Waymo has announced to participants in its Arizona pilot program that fully driverless vehicles will soon be available for rides in a commercial environment.
• Waymo also added that “this is just the next step as we travel down the road to a fully self-driving future” implying that this move is merely a return to what it was doing before.
• Instead of having a driver in the passenger seat, I suspect that Waymo is going back to having a driver in a remote centre for each vehicle who is ready to take over at any point should the vehicle get into difficulty.
• If this is where Waymo is going, it is not autonomous driving in my opinion as the whole point is not to have the attention of a human constantly focused on the driving of the vehicle.
• Yandex is intending to go fully autonomous in 2 parts of Russia this year which it claims will be fully autonomous.
• Yandex has a big press/analyst event in Russia next week showcasing its solution and test centres which I suspect has something to do with the timing of Waymo’s announcement.
• Yandex claims that there will not be a safety driver ready to take over at any point, but staff will be available at a remote-control centre should the vehicle get into trouble.
• However, the areas that Yandex is launching are heavily geofenced and from looking at them on Google maps (see here and here), the street layout is extremely simple and there is almost no traffic.
• Waymo is moving ahead in a more complex environment (Phoenix, AZ) but it too will be heavily geofenced and I suspect there will be remote monitoring at all times.
• I think Waymo is keen to ensure that it is not seen as falling behind Yandex in terms of launching autonomous driving.
• Despite these moves, the reality remains that autonomous driving remains as far away as ever because neither of these systems are close to being able to deal with real-life scenarios.
• This is because some or all of the systems being worked on use deep learning which RFM research has found to be fundamentally flawed in many use cases.
• RFM has determined that deep learning works optimally in use cases where the data set being relied on by the algorithm is both finite and stable.
• The data set of the road is neither of these things which is why the grand ambitions of all players to launch autonomous driving in 2019 and 2020 came to a crashing halt at the end of 2018.
• Hence, I think that the rollout of autonomous vehicles will be incremental as the systems gradually learn to handle increasingly complex environments.
• RFM’s analysis still has Waymo out front with Apple and Uber languishing at the back of the pack.
• Yandex remains the dark horse as it has yet to provide any data for comparative purposes but based on its demos and its AI heritage, I expect it will feature towards the top when it does publish some figures.
• I am hopeful that it might do this at its event next week.
• RFM sticks with is 2028 timeframe for when autonomous driving becomes a commercial reality.
• I still see a lot of consolidation meaning that OEMs do not need to rush to sign up for one solution or another as there will be plenty to choose from and at much cheaper prices in a few years.