After years of development, numerous hurdles, and uncountable unfulfilled promises, it looks like the robotaxi industry is finally taking shape. Over the past couple of years, IDTechEx has reported on the emergence of commercial robotaxi services but has always needed to caveat this with the fact that they have been heavily restricted. There would always be some detail that would make the milestone not quite a mile, like the service being available from 12 am to 6 am or being limited to a very small part of the city. Well, in 2024, the shackles are coming off, and the industry is getting ready for take-off.
The first part of this article looks at where the industry is now and how it has got there. IDTechEx examines this topic through the lens of numbers describing and measuring the industry, ultimately building to a prediction of where the industry will be ten years from now.
Nearly 25 million miles of testing
California has been a key hub of autonomous vehicle development with several factors culminating to form a self-driving haven – an abundance of technical expertise in Silicon Valley, a seemingly endless ability to finance this expensive pursuit, a local government willing to work with the tech companies, suitable road infrastructure, mostly fair weather, the list goes on.
Some have said that robotaxis have been allowed to move too fast, and some have called it a wild-west type environment with innovation running free. But if the robotaxi companies are cowboys, they are certainly well-documented ones. Each robotaxi company has to have correct permits from California DMV, and each mile driven by each car is recorded.
Between 2018 and 2024, the companies testing in California recorded a cumulative 24.7 million miles between them. That’s more than the average human would accumulate in 20 lifetimes. Leading the pack by some way is Waymo, which recorded more than 13 million miles in that period using its fleet of more than 430 vehicles. With over half the miles of all testing in California, it is no surprise that Waymo is one of the top players in the field.
Figure 1. Total robotaxi testing miles in California 2018-2023.
600 collision reports
As robotaxis have become more present in places like San Francisco and cities across China, they have also become more present in the news. Teething issues are accepted or at least tolerated across most new technologies; it is just accepted that things won’t be perfect on the first try, but autonomous driving carries so much additional risk that it is held to very high standards. Whenever a collision or incident happens, the media heavily scrutinizes it. Accidents involving robotaxis range from the bizarre, like the Waymo parking lot orchestra of horns in August 2024, to the frighteningly serious, such as the collision involving a Cruise robotaxi dragging a pedestrian under the vehicle in October 2023.
California DMV keeps records of all collisions involving autonomous vehicles. Between January 2019 and July 2024, 600 reports were submitted. Upon reading these reports, it becomes clear that the roads are not safe, but it is not robotaxis that are to blame. Out of the 600 collision reports that IDTechEx has read, blame could only be attributed to an autonomous driver causing a collision on 29 occasions, meaning that in collisions involving autonomous vehicles, a human is to blame 95% of the time.
Coincidently, 95% is the widely accepted figure for the rate at which humans cause all road accidents, with the remaining 5% attributed to other causes, such as mechanical failures, wild animals jumping onto the road, and freak accidents. This is somewhat oversimplified, though, as in many of the 571 human-caused collisions, the autonomous vehicle wasn’t even driving, but a human test driver was in control at the time of the crash. So, it is necessary to dig a little deeper.
Conducting autonomous testing with no test driver on board has been permitted since 2020. In 2023, robotaxis in California drove 3.3 million miles without a safety driver on board. During this activity, these vehicles recorded 58 collisions, one per ~57,000 miles. However, of these collisions, the autonomous driver was to blame only 11 times; in other words, an autonomous vehicle caused one collision per 300,000 miles in California in 2023. By comparison, IDTechEx research finds that, on average, US drivers are involved in a collision roughly once per 200,000 miles across the country, with this increasing to once per 100,000 miles in the bustling metropolis of San Francisco.
Figure 2. Collision rate comparison.
It is still very difficult to compare human and AI driving proficiency and say that one is certainly better than the other; for a start, which human should an AI be compared to? Anyone who drives on public roads knows that the quality of human driving has a broad spectrum; it wouldn’t be controversial to say that an AI driver is safer than the worst drivers on the road. The numbers suggest that their safety is now comparable to that of the average driver, but surely, they need to be better than average. So, what do the best drivers look like? And how can that be described numerically in a way that autonomous cars can be benchmarked against?
When looking through the crash reports, one thing is certain: humans cause collisions in ways that would be alien to robotaxis. IDTechEx has seen crash reports where humans have hit robotaxis by speeding, running red lights, evading the police, not checking mirrors, and even selecting reverse instead of drive at a set of lights. Human mistakes and failings. Robotaxis, on the other hand, are guilty of causing crashes that seem utterly bizarre to humans. Not being able to correctly interpret an articulated bus, driving through police caution tape into a closed intersection, and groups of robotaxis causing their own phantom traffic jams are all examples of incidents that left people scratching their heads and wondering how the robotaxi got it so wrong. Despite these oddities, it is undeniable that progress is happening.
11 pioneering cities
Robotaxis have been around for a while now. The DARPA Grand Challenge, an autonomous vehicle competition that kickstarted the entire industry, happened some 20 years ago. Since then, cities across the world have become interested in robotaxis. In 2024, IDTechEx recorded nearly 100 unique combinations of robotaxi companies active in different cities. These range from short trials with the technology to full commercial services where customers can pay to ride in a self-driving car with no safety driver present.
China and the US are the hotspots for robotaxi development. Between the two, there are more than 30 cities engaged in robotaxi testing, more than the rest of the world combined. Until 2020, there was very little activity across the rest of the world, but things have picked up since then. Leaders from the US and China have started spreading; Cruise has a presence in the UAE and Japan, while WeRide has conducted some testing and trials in Mongolia. However, the real challenge for all these companies and all these host cities is to progress from a highly controlled and closely supervised trial to some form of commercial release.
Figure 3. Commercial robotaxi services.
At the time of writing this article, ten cities across the US and China have commercial robotaxi services, with Waymo expected to come online in Austin by the end of the year. There are many more cities across the US and China where robotaxi testing is present and commercial releases are possible. Additionally, other countries across the world are getting ever closer to launching commercial services. The UAE has previously set the target of having 4,000 robotaxis in Dubai by 2030. Germany has a commercial autonomous vehicle deployment framework, and IDTechEx expects the first services to be launched soon. The UK has recently passed its Autonomous Vehicles Act, which provides for commercial robotaxi deployment from 2026. This is certainly an exciting time for robotaxis, and IDTechEx is looking forward to what the next 10 years will have in store.
10 years from now
The restraints placed on autonomous cars are beginning to loosen. Cities like San Francisco, Los Angeles, and Wuhan now have wide operational areas for robotaxis, and more cities are coming online each year. IDTechEx believes robotaxis are currently tip-toeing around their hockey-stick moment. Over the next 10 years, one to two new cities with commercial robotaxis per year will become a few new each year, then a handful, growing into a flurry of cities and services coming online, similar to how Uber grew at the start of the 2010s. IDTechEx fully expects that robotaxis will have a global presence, not ubiquitous, but it will be difficult to find a major city in the developed world without some form of commercial robotaxi service.
Robotaxis are not the only form of autonomous vehicle on the roads. Personal vehicles will also start benefitting from more highly automated technologies. In 2035, IDTechEx expects the first consumer SAE Level 4 vehicles to be hitting the market. However, other vehicle sectors are also exciting, particularly autonomous buses and trucks.
Autonomous trucks enter the commercialization phase, proving their economic benefits
The commercialization of autonomous driving technology in the trucking industry has rapidly advanced in 2024, presenting substantial potential to address critical industry challenges such as high operational costs, driver shortages, and safety concerns. It is expected that the global trucking fleet will be supported by an additional half million SAE level 4 (L4) autonomous drivers, while the remaining drivers will either transition into supervisory roles or continue to operate in less developed regions.
IDTechEx’s recent research on over ten companies in the autonomous trucking sector reveals that the industry has officially entered the commercialization phase, marked by deep collaborations across the global supply chain. Observing the market, it is evident that active startups are universally building their software or retrofitting technology on truck chassis provided by Original Equipment Manufacturers (OEMs). This approach not only alleviates the pressures associated with mass production but also allows these startups to leverage OEM technical support. The industry is thus evolving into a combined business model comprising OEMs, downstream logistics companies, and autonomous driving firms.
China has led the way in achieving mass production and commercial delivery of autonomous trucks. Companies like Inceptio and DeepWay have made significant sales (exceeding 1,000 units) in their respective segments—internal combustion engine (ICE) and pure electric autonomous trucks. Notably, after obtaining a license for autonomous driving pilot operations on highways in China, Inceptio has established long-term partnerships with over 13 entities, reaching a milestone of one hundred million kilometers in May this year. In the process, Inceptio has shifted its focus in the Chinese market from dual-driver operations (L4) to single-driver operations (L2+/L3), aiming to reduce costs and increase efficiency. This shift has extended the original transportation route from 600 km to 800-1000 km, while optimizing the driver-to-vehicle ratio to 1.2:1 in long-haul transportation, providing tangible cost reductions for customers.
According to information from key companies interviewed by IDTechEx, this strategy not only reduces driver employment costs by up to 40%, but also offers drivers approximately 8% increase in income, further diminishing China’s demand for truck drivers. This represents one of the two major commercialization paths for autonomous trucks identified in our report. The second path involves establishing high-quality L4 fixed routes, often referred to as “hub-to-hub,” for short-distance (300-700km) autonomous or unmanned transportation. Sweden’s Einride is a notable example of this approach, with three route plans in the UK, Norway, and Sweden. Companies must pre-plan these routes for downstream transportation customers and set up related auxiliary facilities along the way to ensure the smooth operation of L4 autonomous driving. As an electric truck company, Einride has designed an autonomous driving network called “Grids” to optimize fleet management, tractor swaps, and battery management. Einride is already operating independently in Europe and the United States.
IDTechEx believes that at the current stage, L4 autonomous trucks require more efficient route planning to reduce uncertainties along the way, thus improving the quality of the Operational Design Domains (ODD). The L4 hub-to-hub transport can consolidate cargo from multiple clients and maximize the use of the same ODD for L4 level transportation. This approach can save vehicle energy and enhance vehicle utilization.
Autonomous truck market outlook for the next ten years
As policy restrictions on autonomous trucks gradually loosen, early-stage commercial testing is taking place in regions such as Tianjin (China), Texas (USA), the UK, and Sweden, with the formation of relatively stable business models. IDTechEx believes that autonomous trucks have already demonstrated their commercial viability. Over the next decade, similar business models are expected to be validated in more markets. Developments are anticipated across various levels of autonomous driving, including L2-L3 assisted driving routes, L4 hub-to-hub operations, and L4 autonomous ports. Autonomous trucks are expected to gradually penetrate intercontinental road transport, eventually replacing all human-operated segments in transport routes.
Looking at market segmentation, developing countries such as China, with relatively low labor costs, are likely to see market penetration by L2-L3 trucks, moving from dual-driver to single-driver operations while enhancing driving safety and achieving more economical efficiency. In developed regions like the USA and Europe, hub-to-hub operations are expected to lead the way due to more standardized road infrastructure, relatively well-developed regulations, and more controllable road environments. In these regions, electric-powered platforms are anticipated to further optimize costs in short-distance hub-to-hub operations, thanks to their lower energy costs and reduced overall carbon emissions.
IDTechEx predicts that by 2035, driverless missions will spread across the majority of the US, Europe, and China in various operational forms. Truck companies and governments will contribute to the development of road infrastructure for L4-level autonomous trucks.
Roboshuttle and autonomous bus players decline, yet secure open road-testing approvals globally
While autonomous trucks have seen significant progress, the development path for roboshuttles and autonomous buses has been comparatively challenging. Roboshuttles, which are small, fully electric vehicles operating at Level 4 autonomy, were once seen as an ideal solution for last-mile transportation. At their peak, over 25 companies were competing in this space, positioning roboshuttles as a promising mobility solution within the autonomous driving industry. However, IDTechEx has observed a steady decline in the number of active players in this market from 2020 to 2024. Contrary to initial expectations, the industry has not progressed to large-scale commercial testing phases.
The primary factors contributing to this market contraction include inadequate funding, which has prevented many companies from advancing beyond small-scale tests, and a lack of public interest, which is essential for establishing a viable business model in this sector. Several notable examples illustrate the challenges faced in the roboshuttle market. Once a leader in the industry, Navya faced financial and multifaceted issues, leading to its joint acquisition by Gaussin and Macnica and the formation of GAMA. Macnica is expected to fully acquire GAMA, with plans to establish its business model in Japanese mobility markets. ZF has delegated the manufacturing, sales, and operations of its roboshuttles to partners Oceaneering and Beep, ceasing internal development and focusing on supplying the necessary software and hardware for autonomous driving. Cruise suspended its commercial service following a robotaxi incident in October 2023 that resulted in serious injuries and led to the revocation of its autonomous vehicle testing license, halting its roboshuttle project. May Mobility has shifted its business model from using Polaris GEM shuttles to leveraging Toyota Sienna for its robotaxi services.
Similarly, the number of companies actively developing all types of autonomous buses has decreased by 18%, from 21 to 18. Although this decline is less pronounced than in the roboshuttle sector, the industry still has not established a large-scale commercial model. The typical ODD for a bus requires interactions with a broad range of road users, including more pedestrians, cyclists, bus stops, passengers, and surrounding vehicles. The larger size of these buses, coupled with their intricate ODD, presents challenges similar to those faced by roboshuttles. In many forward-looking scenarios, autonomous buses are still viewed as fundamental urban infrastructure, which implies the necessity of substantial government support to advance these projects. This need for government involvement is one of the main reasons for the relatively slow progress in the development of autonomous buses. Among the three types of autonomous buses—minibuses, midibuses, and city buses—the number of companies working on city buses has decreased the most. In contrast, there has been a slight increase in the number of players focusing on minibuses, driven by the limited operational scenarios for city buses and the entry of some roboshuttle developers into the minibus market.
Despite the overall downturn and all the challenges faced, some players achieved significant milestones between 2022 and 2024. In August 2023, WeRide received the M1 and T1 licenses from Singapore’s Land Transport Authority (LTA), allowing their Robobus to undergo large-scale public road tests. By July 2024, WeRide launched Robobus shuttle services in Sentosa, marking Singapore’s first publicly accessible L4 autonomous Robobus route and the industry’s first truly commercialized Roboshuttle operation. Similarly, China’s QCraft is conducting open tests in 10 cities across mainland China. Pix Moving is deploying 100 roboshuttles for testing in 16 cities worldwide, leveraging its production line to support these trials. These achievements point to a direction for the early commercialization of roboshuttles, namely providing services in limited areas with high-quality infrastructure, highly controlled access, and short-distance journeys.
In parallel, two city bus projects are still moving forward with commercialization, supported by strong backing from various stakeholders including government. For example, in Seoul, Hyundai launched two L4 ADS (Automated Driving System) night bus services in December 2023, operating on a 9.8 km route from Hapjeong Station to Dongdaemun Station. This service has transitioned to a paid service in the first half of 2024, representing a huge milestone for the autonomous vehicle industry. The project benefits from the support of the local government, highlighting the role of collaboration in advancing such initiatives. In the UK, a longer autonomous bus route is in operation—a 14-mile stretch between the Ferrytoll park-and-ride in Fife and Edinburgh Park. This service runs frequently and can accommodate around 10,000 passenger journeys per week, demonstrating the potential of autonomous buses in public transportation.
The minibus sector has experienced a surge in activity. This growth is partly driven by support from local governments and the shift of some roboshuttle developers toward this faster, less commercially constrained market. Several minibus projects are now targeting specific commercial applications. For example, VW‘s MOIA project in Germany focuses on city center coverage, while Aurrigo in the UK and Singapore is involved in passenger transport and cargo operations at airports.
These commercial breakthroughs underscore the crucial role of government support in facilitating on-road vehicle testing in 2024. However, with the rapid development of robotaxis and other on-demand mobility services, the uniqueness and value proposition of roboshuttles and autonomous buses are increasingly coming into question. The expectation of recouping costs through fare revenues alone is also proving to be unrealistic in the short term. IDTechEx’s report provides insights into early commercialization, suggesting that sustainable progress will depend on the additional value that roboshuttle and autonomous bus services can bring, such as addressing commuting gaps within regional areas, to attract continued investment from governments and customers. IDTechEx believes that during the early commercialization process, companies should focus on vehicle delivery and maintenance services to secure their hard-won market share.
Autonomous bus and roboshuttle market outlook for the next ten years
Autonomous buses and roboshuttles are currently undergoing near-commercialization testing in cities worldwide, including Singapore, France, China, and South Korea, with tests involving real passengers in certain areas and managed zones. IDTechEx believes that over the next decade, autonomous buses and roboshuttles will still need to prove their commercial value to attract continuous investment and government support for market development and policy endorsement.
Each has its challenges: roboshuttles need to overcome the uncertainties posed by dynamic routes and the performance challenges of navigating narrow urban streets. Testing in such areas, closer to residential zones, means these vehicles will be under greater public scrutiny. Buses face the challenge of maintaining relatively high speeds while ensuring the safety of dozens of passengers, which is crucial for gaining public trust.
Over the next ten years, the number of cities deploying roboshuttles is expected to increase by one or two per year, initially focusing on pilot zones similar to those in Singapore, with later integration into urban public transportation networks as a last-mile solution. Autonomous bus pilots will continue to be primarily government-supported projects, such as Hyundai’s operation of nighttime autonomous shuttles in South Korea. However, on-demand autonomous services are growing rapidly, with robotaxi services already operating in regions of the US and China.
On-demand mobility solutions more accurately address current market needs, though IDTechEx believes that roboshuttles and buses still have the potential to capture specific market niches. By 2035, roboshuttles are expected to be deployed in major cities and operate without the need for safety drivers, while buses likely to continue requiring safety driver supervision in the near term.
For a full analysis of the autonomous mobility landscape, please refer to the portfolio of IDTechEx research on the industry, with the full list of drill-down reports available at www.IDTechEx.com/Research/Autonomy.
Dr. James Jeffs
Principal Technology Analyst, IDTechEx
Technology Innovations Outlook 2025-2035
This article is from “Technology Innovations Outlook 2025-2035”, a free collection of insights from industry experts highlighting key technology innovation trends shaping the next decade. You can download the full collection here.
About IDTechEx
IDTechEx provides trusted independent research on emerging technologies and their markets. Since 1999, we have been helping our clients to understand new technologies, their supply chains, market requirements, opportunities and forecasts. For more information, contact [email protected] or visit www.IDTechEx.com.
