Autonomous cars – here, now, no safety driver!
There are many challenges to the implementation of fully autonomous cars, able to operate totally without a driver on public roads, so the announcement last week that autox – a venture backed by Chinese e-commerce giant Alibaba had launched a driverless robotaxi on the streets of Shenzhen in Southern China was significant. (See a video here). They are not the first to take this step, as Waymo took the safety driver out of some of their cars in downtown Phoenix last October, allowing operations to restart during the pandemic without health concerns. That in itself is ironic – that the perceived risk to life and limb from infection was seen to be greater than that from the driverless car…
We have always been sceptical in ICDP about the potential for truly driverless cars – Level 5 functions in the SAE terminology, compared to Level 3 as the most advanced currently available in a production car. There are a number of reasons for this, including human factors associated with the response time of an individual to react in the event of a system malfunction (assumed in Level 4) and the remaining technical challenges of making a Level 5 system work in all weather conditions on a public road network. The human factors cannot be addressed unless we could somehow reprogramme the human brain to maintain situational awareness in parallel to whatever they are doing on their autonomous car journey, so the head of the Volvo autonomous car programme has indicated that this stage is not a realistic proposition, saying that the human is the worst possible fallback system for an automated system. Although the autox and Waymo driverless cars are connected to a central operator who can intervene, this is in the event of a situation where the car’s systems have become disorientated for some reason and it has come to a halt, waiting for help from the remote operator.
This disorientation is not unusual. There is an interesting video on the Waymo blog by a rider called Xavier who describes the confusion that the Waymo car experiences when trying to manoeuvre in a busy shopping mall parking lot, with lots of jerky stop-start movements as it responds to people, cars and shopping trollies. John Krafcik, the very capable and realistic CEO of Waymo, has described how they have had to programme the cars to inch forward at the four-way give way junctions that are typical in many US suburbs, sensing whether the other cars will give it priority, rather than just sit there indefinitely until there are no other cars at the junction.
Krafcik has also been honest about the technical challenges of making the autonomous systems work in all weather conditions. Even the most sophisticated LIDAR systems are blinded in snow and heavy rain, which is why these first driverless launches have been in the much sunnier streets of Phoenix and Shenzhen, rather than Manchester or Brussels. It is also highly dependent on a well-mapped and relatively open road infrastructure. Again, if you look at the road and traffic conditions on either of the videos, there are wide open highways with few parked cars or small side streets – far from the typical European urban environment. Even then, the Waymo cars are restricted to a twenty square mile area of Phoenix called Chandler which has been mapped in minute detail by Waymo
The main driver of these driverless projects (if you will excuse the confusing terminology) is not to give the average driver the chance for a rest on their daily commute, but to eliminate the cost of the driver from taxi services and freight haulage. Frost & Sullivan have estimated that the driver accounts for 80% of the total per mile cost for a ride-sharing service like Uber, and their only route to profit is to eliminate the driver, which is why they have invested heavily in their in-house autonomous car project, Advanced Technologies Group. However, even then, their ambitions are modest, anticipating a “limited geographical capability”, i.e. not London or Paris.
So, do these two driverless robo-taxi launches change my mind about the future of autonomous cars? The short and clear answer is “no”. The technology will continue to advance, and I’m looking for an excuse to travel to Shenzhen or Phoenix to have a go, but as something that will directly affect the demand for, or usage of, cars as a form of personal mobility, the technical, regulatory and human barriers remain high. The technologies will affect all cars, as we have already seen with current driver assistance system (ADAS) capabilities, but I’m not expecting a robotaxi to turn up to drive me from my home in my lifetime – unless I move.