Reading time: 6 minutes
But what is he doing? No way! Later, the cyclist would describe the van’s driving behaviour as unpredictable. “The car was driving in the left lane with the indicators signalling right,” he recalls, “but it didn’t move over, instead he stopped repeatedly and was holding up the cars behind him.”
The cyclist also hosts a podcast, where he reported: “In the end, he signalled left and stopped on the right. That was strange. I don’t know if I noticed it during or after the incident, but there was a sign on the van that said ‘Self Driving Vehicle’.”
David Gölnitz is the spokesperson for MOIA, a technology company that has been operating a fleet of self-driving shared shuttles for test drives in Hamburg since November 2024. The autonomous VW ID. Buzz AD electric buses can drive at autonomy level 4 within a designated area of 37 square kilometres in the Hamburg-Nord and Wandsbek districts. However, a safety driver is still required to be on board before they can attain approval for regular operation.
The software learns from mistakes
Gölnitz reacts calmly to the reporter’s story. “This happens all the time,” he says. “And that’s exactly how it should be.” When the vehicles are driving on the right, it is often due to the solid centre line. “A private person might glance and cross the line. An autonomous car always strictly adheres to road traffic regulations.” The car sends a request with various options for action to the control centre, where an operator can give the go-ahead. Until this is granted, the vehicle goes into ‘minimum risk mode’ and stops.
This can also happen when the 13 cameras that capture the surroundings in real time register situations that are difficult to interpret, explains Gölnitz: “For example, pedestrians walking briskly towards a red traffic light with no sign of slowing down.”
“It’s extremely important that we have cases like this,” said the MOIA spokesperson, referring to the algorithm-based logic of autonomous driving. “The more data we have, the faster the software will be able to respond. Our vehicles are still at the very beginning of their training.” In theory, software only makes a mistake once. Then it has learned. Much unlike humans. Real-world projects are crucial in ensuring that autonomous driving is introduced safely. It is only in this environment that regulations can be applied and further developed. Studies conducted on isolated test sites no longer reflect the complexity of road traffic. Therefore, legislation and approvals only progress through practical testing, whether in public transport, logistics or urban traffic. And while Hamburg can serve as a blueprint, other cities have different requirements. And what remains crucial is the cooperation between all stakeholders, from technology providers and authorities to mobility services.
“Of course, we will only be accepted on the road if we fit into the flow of traffic,” says Gölnitz. Market research and acceptance studies on the Hamburg project are still ongoing, but since its inception in 2023, there have been no more than a dozen complaints from irate motorists.
The cities of Germany have long since started a competition: Darmstadt, Friedrichshafen, Ilmenau, Karlsruhe, Mannheim, Monheim am Rhein, Munich, and Waiblingen all have their own projects. In Hamburg, MOIA and Hochbahn are joining forces under the ALIKE research project with their VW ID. Buzz AD and Holon vehicles.
MOIA spokesperson Gölnitz says that public transport is just one of several ways to establish autonomous driving: “We are adapting to market conditions.” He also announced plans to introduce ride-pooling or ride-hailing systems in North America in collaboration with Uber and the Volkswagen Group of America. The ride-sharing provider Uber, which initially had a rather low-key start in Hamburg, is now becoming a global market player under the wing of VW.
The autonomous MOIA is just a small step away
The people of Hamburg are proud of their shiny bronze shuttles, which will be able, in the long term, to operate without drivers in a defined and data-screened area. Regular Level 4 autonomous operation is set to launch in 2027, initially for a specific group of people that anyone can apply to be a part of.
Hamburg’s transport senator, Anjes Tjarks, has set a target of putting 10,000 autonomous shuttles on the roads by 2030. The city believes that a ‘cascade effect’, i.e. a rapidly growing impact on urban transport – for example, through improved accessibility to public transport services – can be achieved with just 2,000 autonomous vehicles.
Thanks to MOIA’s close collaboration with Hochbahn and the public transport company HVV, the autonomous fleet in Hamburg is a valuable addition to the city’s mobility. Hamburg’s train services too are planning to achieve significantly higher frequencies on the S-Bahn thanks to shorter, autonomous trains.
Autonomous driving will be used primarily in combination with sharing models initially; and MOIA has a head-start thanks to its existing driver-operated ride-sharing service.
MOIA, this hybrid form of bus and taxi service, creates a denser public transport network, integrates young people and adults, and is increasingly reaching rural areas. It reduces accidents and the number of cars on the road. This increases space in the city and improves quality of life, while the quiet, emission-free vehicles are also more energy-efficient than taxis thanks to optimised routing and higher utilisation rates. There are so many advantages that it is difficult to decide which to mention first.
Autonomous driving is becoming a global business
The market is growing rapidly, driven by advances in AI technology and falling software costs. According to a Goldman Sachs study from 2024, there could be several million autonomous vehicles on the road worldwide by 2030. This would create a robot taxi market worth over $25 billion, even if autonomous vehicles only account for a small proportion of the global vehicle fleet.
The growth is also impacting other sectors, according to the study: “The acceleration of the spread of autonomous vehicles could boost a number of stocks in a range of sectors, including chip manufacturers, ride-sharing companies, developers of autonomous driving technologies, and car manufacturers.”
In the US and China, where robotaxis have been available for some time, the market is dominated by tech companies such as Waymo, Uber, and Baidu.
Pragmatic Europeans do not require Level 5
Level 5, which is necessary for fully automated driving, is still a long way off, at least by European standards. Europe prioritises safety over speed, having published a set of regulations in 2022 that enable regular operation at SAE Level 4. While progress in the US and other countries is often made through extensive road testing, here the emphasis is on ensuring that autonomous systems are not only equivalent to, but better than, human drivers. The complexity of traffic situations is a particularly crucial factor in this regard.
David Gölnitz from MOIA considers fully autonomous driving to be ‘not worth persuing’ at the moment. “With Level 4, we have approval for certain operating areas, which can be expanded as desired once the necessary data has been collected. Even an entire city could become an operating area. The ideal application for autonomous driving is Level 4 in local transport or shared mobility.” But what about the industry’s great utopia: the elusive Level 5? Gölnitz says, “That’s a big step with little benefit.” He justifies this view by citing the current framework conditions: from today’s perspective, the costs of autonomous private transport are too high, and it would not ease traffic congestion in cities. Traffic jams and parking problems would remain.
Challenges of acceptance
In addition to the technical challenges, the question of public acceptance remains. Although autonomous vehicles offer greater safety and new mobility options, the general public is still cautious. Transparent testing and evaluation procedures are needed to demonstrate how the technology works and what benefits it offers, in order to build trust. As an additional challenge, not everyone is comfortable with sharing rides with strangers. Although a MOIA can accommodate up to six people, the evaluation found that the vehicles are not always booked to their full capacity in everyday use. For the concept to succeed in the long term, it must prove that it’s economically viable and affordable.
