GM Cruise – Speedbump

GM Cruise accepts reality

• Autonomous Cruise vehicles are no longer expected to be on the streets of San Francisco this year as Cruise has finally accepted that this task is far more difficult than it thought to crack.
• This is exactly in line with RFM’s expectations where it’s AI research (see here) clearly indicated that Cruise’s 2019 expectations were unrealistic.
• In very careful wording Cruise (see here) has said that an increase in testing and validation will have “the effect of carrying the timing of fully driverless deployment beyond the end of the year”, which in English means “We don’t know when it will be ready. Please stay tuned”.
• Cruise stubbornly held on when almost everyone else pushed back their expectations of commercial autonomy to “sometime in the next 10 years” insisting that it would be ready this year.
• I always thought that this was an odd position to take because its challenge is actually more difficult than almost all of the others as it is aiming to deploy in San Francisco which is a dense urban environment.
• All autonomous driving systems have a heavy deep learning component to them which has both great advantages as well as disadvantages.
• The advantages are that the system can learn and be in a position to deal effectively deal with situations that it has previously encountered without being explicitly programmed for every eventuality.
• However, these systems only work properly in environments where the data set is both finite and stable (see here).
• When it comes to the road, the data set of a dense urban area is less finite and less stable than an open grid with no traffic or the highway.
• Hence, it is much more difficult to get autonomous vehicles working reliably in dense urban areas than in the countryside or on highways.
• Mumbai or Cairo are likely to some of the most difficult to attempt due to their density of traffic and the level of chaos.
• With Cruise now in line with everyone else, only Tesla still reckons to go commercial this year or next which I have already addressed (see here).
• Yandex also is intending to go live with a commercial trial this year but the parameters for this are very rigid (see here).
• Despite this, I am happy to keep Cruise as the No. 2 in autonomous driving behind Waymo with Yandex, which has yet to release data, as my dark horse of the industry.
• Uber and Apple are so far away from having anything remotely viable that I continue to think that both will have to make acquisitions if they wish to continue.
• Zoox or Nuro are where I would look first.