Dear Reader,

In January 2023, Mobileye received a recommendation for authorization from the TÜV (German technical certification institute) for its "Robo-Taxis" in Germany. Field tests are planned for Munich and Darmstadt but so far, no vehicle has been spotted in operation. It is bound to be exciting, particularly in Munich, as in my experience driver's behavior there is sufficiently ruthless and egoistical for a true acid test. Munich is also currently, and no doubt in the future too, a hotspot for blockades by the "last generation".

At the moment, a closely watched field test by Cruise and Waymo is being run in San Francisco, despite (fundamentally non-spectacular) incidents (Cruise drives in wet cement, functional blockade by unknown reason, spontaneous swarm formation of several Cruise vehicles in the same junction with resulting mutual blockade). Even first response vehicles are sometimes not recognized and thick fog of course is an obstacle too. Most of the events are not serious. However, and this is commented on by experts, the argument that the faults are not serious is a poor one in terms of safety. As long as the message is propagated that the machine drives better than a human, it cannot be allowed to make any mistakes at all. Therefore, both the road works sign and the ambulance have to be registered, since the computer does not daydream nor listen to distractingly loud music.

So there are new, or at least other mistakes appearing that the human does not make: The aforementioned blockades, for example, were caused by loss of contact with the company server. The event could thus be seen as a proof-of-concept, as if the communication link is lost, the systems enter a safe state and the vehicles do not represent any danger. However, extrapolating from the intended function, this is a meltdown if everything comes to a standstill in the rush hour. Wherever there are also blockade situations without loss of contact, where the AI fails and cannot find any further option, any explanation why the event happened indicates a useful insight. However, experts have their doubts that the root causes are found since some faults repeat themselves.

Either way, there is another threat on the horizon: An activist group in San Francisco calling themselves "Safe Street Rebels" plays trap the Cap by placing traffic cones on the hood of autonomous vehicles, which locks them thus necessitating local, manual unlocking. This would give the activists an alternative to gluing themselves to the road and provide them with a further method of blockade: practically fully automatic.

Enjoy reading this edition.

figure 1

Robert Unseld

Responsible Editor