Autonomous Cars: A Technological Revolution in Slow Motion

Autonomous passenger cars are seen as the holy grail of AI-powered self-driving technology, but they are also the most complex use case. This is why robotaxis are moving on thin ice. So, let's have a look beyond that slippery icy surface.

Robotaxi operator Cruise managed to make booking autonomous cab rides "hip" in San Francisco last year, only to see its license revoked after a traffic accident. Also, Elon Musk's presentation of Tesla robotaxis was postponed to October. To be fair, manufacturing robotaxis is an incredibly difficult job, because the public tends to take every accident as definitive proof that autonomous cars aren't "ready for prime time" yet – not matter how many accidents are caused by human drivers under similar circumstances.

At the same time, the internet is full of videos showing Cruise, Waymo, and other autonomous cars failing to work their way around ordinary traffic obstacles. It's a technological revolution happening in slow motion, in an arena surrounded by billions of smartphones with video cameras.

But let's not make the mistake of assuming that the autonomous vehicle revolution won't happen, or will only do so in a distant future. There are numerous other settings where autonomous driving already provides real value – and where it can be deployed much faster than in busy city streets. Autonomous vehicles are already succeeding – just not yet in the heavily scrutinized arena of the taxi business.

Autopilot Mode Working Effectively in Many Industries

It's not a coincidence that the term "autopilot" – boldly, if somewhat incorrectly, appropriated by Elon Musk for Tesla's full self-driving mode – comes from the aviation industry. Compared to city traffic, the airline industry's airways are a calm, clearly defined, controlled environment – and this is where autonomous vehicle technology really shines.

But one doesn't have to go miles above ground to find settings where autonomous vehicles can prove they are "ready for take off" – or rather, ready to make vehicle operation smoother, safer, and less costly. Just think of autonomous forklifts in warehouses; autonomous trains and subways; autonomous trolleys moving freight through tunnels; autonomous heavy-duty cranes at cargo ports; or autonomous long-distance trucking on standard highway routes between logistics hubs.  

High Potential Comes with High Demands

In these environments, vehicle operation and traffic tend to be tightly controlled, often sealed off completely from the general public. So operators don't have to worry about children suddenly running onto the street, or pedestrians crossing the vehicle's path without watching out for oncoming vehicles (because being busy with their smart phones).

That doesn't mean that there aren't any challenges. However, these challenges are mainly technical. After all, modern vehicles are basically data centers on wheels. They combine numerous IT systems, from ADAS (advanced driver assistance system) to on-board entertainment, from the dashcam system to the computer that controls EV charging. Accordingly, the compute and storage components must be designed for harsh weather conditions, very high and low temperatures, the shocks of rough roads – and the safety requirements of an industry in which customers’ health is at risk if certain mechanical, electric, or IT components fail.

This is why these components must meet very strict Functional Safety for Road Vehicles standards – up to the highest level of automotive hazards, such as the brakes on all four wheels failing at the same time. These specifications require components to work in a temperature range that most data center servers wouldn't survive – and this applies to the compute components as well as the storage components which are so critical for real-time data management in an "autonomous data center on wheels". At the same time, there is an additional challenge that is never encountered in a traditional data center: it must be possible to provide the over-the-air update functionality to keep modern cars operating reliably and securely.

Beyond technical product specifications, car industry suppliers also have to meet challenging criteria on the manufacturing process level. The automotive industry has established very demanding production part approval processes, allowing them to evaluate whether a supplier meets their high standards regarding manufacturing capabilities, quality control, and production consistency.

These processes apply to the mechanical and electrical components designed for vehicles moving at very high speed, but also to the compute and storage components. After all, reliable compute and storage are a core feature in modern vehicles – even if not core part of the awareness campaigns – as they provide their consistent, smooth, safe, increasingly AI-assisted, and ultimately autonomous operation.

Paving the Way for Autonomous Vehicles

Autonomous passenger cars such as robotaxis are the most visible and therefore most heavily debated variation of autonomous vehicles. But they are an outlier, because they have to operate in the complex and often chaotic environment of busy city streets. In more controlled settings, the question is not whether autonomous vehicles will arrive, but only when this will be the case. That’s why technology providers, manufacturers, and regulative authorities need to cooperate to move autonomous vehicles off the thin ice, and to create a smooth way forward for this mobility revolution, be it on the road, on rails, in harbors, or in other cutting-edge settings. And one day, maybe even in the taxi business.