EE Times Europe – Autonomous Vehicles: How Is Europe Doing?
Advertisement Even discounting the overly optimistic expectations of the recent past, it's clear that autonomous vehicles will have a major impact on the automotive and transportation industries. The AV era is not a question of if but of when, and the "when" varies by region and by AV use case.
This article will focus on how Europe is doing in AVs across multiple segments of the AV industry. Comparisons with the U.S. and China are included where data is available. To get a basis for the comparisons, I have included a big picture of how the AV industry is likely to evolve in the next decade or so.
This includes AV hardware and software technology along with the use cases, standards, and regulations shaping the AV market. Table 1 summarizes the phases of AVs' evolution. The timeline across the bottom doesn't indicate specific years but starts around 2015 on the left and ends in the 2030 timeframe on the right.
More explanations of key importance are given in the paragraphs that follow.
ODD = operational design domain; OEDR = object event detection and response; DDT = dynamic driving tasks; ADS = automated driving system; LSAD = low-speed autonomous driving; MaaS = mobility-as-a-service (Source: Egil Juliussen, October 2022)
Hardware and software
The leading hardware and software companies entered the development phase between 2015 and 2018. They are now deep in the testing phase, which has taken longer than expected. The leaders have done extensive AV modeling and simulation, tallied in billions of miles of virtual driving.
On-road testing is also wide-ranging and is counted in the tens of millions of miles. The leading software players are developing and testing their own software "drivers," which may span multiple use cases. The robotaxis/robotrucks combination is the most common.
The key to moving to the deployment phase is to have the software driver learn as many edge cases as possible. An edge case is an unfamiliar driving situation that the software driver has not experienced in earlier testing. In the deployment phase, hardware system performance, including LiDAR and other sensor performance, is a key factor.
Hardware redundancy is required as a safeguard in the event of system or component failure. Software deployment means driverless operation, which will require teleoperation oversight by a remote operator. The software leaders will provide software driver platforms that can be used by multiple companies across several use cases.
The volume deployment phase will see rapid decline in AV system prices, especially for LiDAR, which is currently the most expensive hardware segment.
The AV use cases have a large range of driving complexity, with fixed-route AVs being the easiest. The sidewalk AV, which is part of the goods-only segment, is also relatively easy because of the vehicles' low-speed operation -- essentially pedestrian walking speed. Robotaxis constitute the hardest segment because of the complex driving environments, including bicycles, pedestrians, and other unexpected road users.
Robotrucks present similar complexity in city driving. Robotaxi highway driving is simpler, but the high speed requires extra safety assurances because of the long distances required to stop a moving vehicle that has high momentum. Extensive pilot operations are now under way, and leading robotruck and robotaxi companies have generated revenue streams from their platforms.
Most of the pilot vehicles in use have a safety driver, though some driverless robotaxi pilots are in operation in the U.S. and in China. The next step is the deployment of specific AV use cases. This phase is likely to emerge from 2023 to 2025, with the timeframe varying by country and use case.
In the U.S., robotaxis and robotrucks are likely to lead. In China, robotaxis will lead. In Europe, fixed-route AVs will lead.
In the volume-deployment era, the robotaxi use case is expected to be the most valuable segment. Robotrucks and goods-only AVs will likely be the second most valuable segment. Mobility-as-a-service (MaaS), covering fixed-/flexible-route use cases, will create synergy with mass-transit operation.
The first standards were the explanations of AV levels and the functional definition of AV operations by SAE International.
SAE has defined six AV levels (L0 to L5), with L0 denoting no autonomy, L1 and L2 denoting ADAS functionality, and L3 through L5 providing increasing autonomy. This article is focused on L4, because L5 is not expected to be viable for a decade or so. The functional definitions of AV operations are important to understand, as they are used across the AV industry, even in the emerging AV regulations.
This is an SAE terminology summary:
o Operational design domains (ODDs) define where AVs can operate and specify the use cases, including many types of passengers and goods transportation. ODDs will have different specifications for each use case and driving segment.
o Object and event detection and response (OEDR) operations monitor and respond to the driving environment via vision and software driver platforms.
o Dynamic driving tasks (DDTs) are tasks performed by a software driver platform, such as steering, speed control, and braking. DDTs also include the OEDR functions and the most important function: crash avoidance.
o The automated driving system (ADS) comprises the hardware and software platforms that perform the entire DDT on a sustained basis for a specific ODD.
The ADS consists of the AV computer system, including the AV sensor system and the software driver platform. As Table 1 indicates, there are already three important AV standards: ISO 21448 (SOTIF), UL 4600, and IEEE P2851. UNECE WP.29 cybersecurity and OTA software updates are also crucial to AVs and are included.
The next level of standards will cover AV use cases. ISO 22737 Low-Speed Autonomous Driving (LSAD) is the first of these. ODDs probably require a standards effort, and work is indeed under way in Europe.
Regulation always lags new technology development, and this is the case for AVs.
California is an exception; the state started managing a test permit program in 2015. These are far from being AV regulations, but they have allowed many companies to start AV testing with a safety driver. Nearly 100 companies have tested AVs in California under this program, which continues today and has expanded to include commercial-use permits.
Currently, 50 companies, five of them European, have permits for testing with safety drivers, and seven companies have driverless testing permits in specific areas in California. Three companies -- Cruise, Nuro, and Waymo -- have deployment permits, albeit in specific areas only. The EU has released the most comprehensive AV regulation, which is currently in draft phase.
This draft covers two key topics: the AV performance specification and the compliance specifications to meet the performance regulations. This legislation will guide the AV deployment phase in Europe and will have direct and indirect impact on other regions. (Editor's Note: EE Times has covered California's AV testing results and the EU draft regulation in detail.) The likely next step in AV regulation is to keep track of how the AV software driver improves as volume deployment expands.
It is especially important to verify that the software drivers are better than human drivers in all use cases -- including how much better, based on crash statistics.
Europe's AV status
In the context of the AV phases described above, how is Europe is doing in developing its AV segments? Table 2 summarizes my perspectives and public information on the global AV industry to offer an overview of Europe's standing in AVs.
AV hardware tech
Europe is behind in most of the hardware categories. This is not expected to be a problem, however, as the continent's auto OEMs can buy what they need from Tier 1s and the supply chain. Infineon, NXP, and STMicroelectronics do well in traditional radar chips.
Tier 1s are strong in traditional radar systems for ADAS and will need to move into 4D radar as that technology grows in importance. Tier 1s also do well in camera systems because of their success in ADAS. Some OEMs may also design their own SoCs for AVs rather than use Nvidia chips.
Volkswagen announced it has a deal with Qualcomm for such a chip. The leading Tier 1s have strong agreements with Nvidia for GPU-based chips and currently provide such AV systems for their OEMs. LiDAR remains a weakness, but several European OEMs and Tier 1s have invested in or have relationships with LiDAR startups.
The leading EU Tier 1s are or will be LiDAR suppliers.
AV software tech
The software driver is arguably the hardest software development project ever -- or at least Waymo thinks so. Europe has a few startups in the AV software driver race, but none are among the leaders. Finland's Sensible 4 is intriguing because the company is focusing on AV software testing in winter weather.
The U.K.'s Oxbotica is focusing on specialty vehicles in industrial segments such as mining and energy. It is also focusing on fixed-route AVs, working with ZF, which is an investor in Oxbotica and a developer of AVs. European automotive companies are expected to rely primarily on U.S.
AV software companies and possibly to work with Chinese AV software companies in China's market.
AV use cases
Europe has been a leader in fixed-route AV use cases, thanks to the availability of autonomous vans from EasyMile and Navya. Both companies have conducted more than 100 tests across many countries and cities on multiple continents. The cost of these AVs, however, is too high for volume deployment.
The entry of EU Tier 1 suppliers such as Bosch, Continental, and ZF could accelerate the fixed-route AV segment. Europe is also doing well in the physically smallest segment: sidewalk AVs. This is due to the success of Starship Technologies, whose vehicles have made more than 3.5 million deliveries and driven more than 4 million miles worldwide.
I consider Starship a European company because it was founded in Estonia in 2014. It is now headquartered in San Francisco but maintains engineering operations in Estonia. In the other use cases, Europe is behind -- especially in robotaxis, which are expected to be the largest future AV market segment.
The difficulty of getting AV and robotaxi testing permits is the leading reason for low European development activity. Further, smartphone-based ride hailing is limited in Europe because of regulatory hurdles in many European cities and countries. Additionally, the high quality of mass transit has limited the need for ride hailing and may similarly slow the growth of robotaxi usage.
Nevertheless, it is likely that multiple European OEMs will enter the robotaxi segment as leading software developers improve their software driver platforms. VW is closely involved with Argo.ai, which started driverless testing in May for ride hailing and goods delivery in two U.S. cities: Austin (Texas) and Miami. Argo.ai has also tested its AVs in Munich with safety drivers.
VW plans to introduce an AV version of the ID.Buzz in the 2025-2026 timeframe. Europe also lags in autonomous trucks, but several EU truck OEMs have relationships with leading AV software driver companies. Daimler is working with Waymo, and VW's Traton subsidiary is working with TuSimple.
Volvo is working with Aurora, and Iveco has a relationship with Plus.
Europe is behind the U.S. and China in AV testing, and the pattern here loosely tracks the continent's status in the various use cases. There has been limited robotaxi testing in Europe. Many of the European automotive companies do or did AV testing in California but generally have had poor results compared with Waymo, Cruise, and some of the Chinese companies.
VW and Mobileye have recently started AV testing in Germany, and Oxbotica has done AV testing in England. Likewise, there is little autonomous-truck testing in Europe. Robotruck testing in the U.S. is done by the top logistics companies and truck OEMs in cooperation with the software driver companies.
Some of the logistics companies use trucks from the leading European truck companies. This is a boon for those companies, as some of their truck models are already interfaced to AV hardware and use the software driver platforms. Iveco recently completed closed-course robotruck testing with Plus in Italy and will start public road testing soon.
Einride is also testing its electrical trucks in Sweden but only in closed environments. Goods-only AV testing is led by Nuro in the U.S., with multiple commercial pilots under way. China's Neolix has a similar product that is the object of extensive testing and pilot operations.
Europe has been the leader in testing fixed-route AVs because of the worldwide efforts of EasyMile and Navya. Oxbotica is also testing fixed-route AVs. Starship, the frontrunner in deploying sidewalk AVs, not surprisingly leads in sidewalk testing.
Much of the activity is on university campuses, with AVs delivering fast food to students. Grocery deliveries are also popular in the U.K. and other European countries.
Europe is very active in developing and releasing technical standards for AV segments and will remain a leader in future efforts. Indeed, Europe is the leader in safety standards.
Much of it stems from the type-approval procedure, which requires a systematic approach to car development; this becomes even more important for AV use cases. Europe has been the driving force to get automotive cybersecurity and remote software upgrade standards on the books, and again, this is even more important for AVs than for current software-defined vehicles.
Europe is far ahead of the U.S. in meaningful AV regulation. Both the U.K. and EU have released solid preliminary AV regulations.
Germany and France have also introduced early AV regulations and are expected to follow the latest EU regulations. The core of the EU regulations is type approval of AVs, which will build growing expertise for both the auto industry and the regulators. In terms of initial safety, the self-certification system used by OEMs in the U.S. is inferior to the European type-approval system.
Fortunately, self-certification has a counterforce: the pattern of aggressive litigation in the U.S. An OEM can expect to be sued when flaws are discovered. But lawsuits take time and are filed in response to a harm that has already occurred, not to prevent a death or injury from happening in the first place.
The AV strategy varies by OEM.
The hardware will be available from multiple Tier 1 suppliers and the emerging AV supply chain. Some OEMs have invested in sensor startups that are pursuing LiDAR and 4D radar development. An OEM's AV software strategy is critical to its success.
A few OEMs have acquired or own a large portion of a software driver startup. GM bought Cruise, and Honda has a Cruise investment. Ford and VW own most of Argo.ai. (Editor's note: See References for links to EE Times' coverage of OEMs' relationships with AV software driver companies.)
Renault, Stellantis, and Volvo have relationships with Waymo. Mercedes-Benz is working with Nvidia on software-defined vehicles, including AVs. Most OEMs are expected to participate in the robotaxi segment, which will eventually lead to personal AVs, probably five years after robotaxis.
Many OEMs will also likely supply AVs to goods-delivery operators and probably will target vans for fixed-/flexible-route AVs. In the testing areas, the European OEMs are lagging the leading U.S. companies. This may be a disadvantage if the robotaxi segment moves to deployment in 2023 and 2024, but probably not for the personal AV market.
It looks like more AV testing is on the way for EU OEMs.
The Tier 1 suppliers will do well in the AV hardware segment. Most are established in the ADAS market and will leverage that experience to enter the AV market. The leading European Tier 1s are well positioned, and some have invested in the LiDAR and 4D radar startups.
The AV software market is more difficult for most Tier 1s. Because many OEMs are working to move more software in-house, it will be challenging for the Tier 1s to build or expand their software businesses, and an AV software play would be particularly difficult. One strategy would be to invest in AV software companies, as some have done.
They could potentially even acquire AV startups, as the funding climate for AVs has gotten tougher. An example is Bosch, which bought U.K.-based AV startup Five.ai in April 2022. Bosch also has an investment in Momenta, a leading Chinese AV software company.
Aptiv is well positioned because it acquired two early AV startups -- Ottomatika and Nutonomy -- in 2016. Those acquisitions became the foundation of Motional, a joint venture between Aptiv and Hyundai.
There is little doubt that Europe is behind the U.S. and China in developing AVs. The key to AVs is software technology, including AI-based neural-network and deep-learning technology, and Europe is behind in those software segments, which impacts European AV technology development.
However, many European automotive companies have strong relationships with key AV software platform companies. A good example is VW, which is a key investor in Argo.ai. Europe is far behind in robotaxis because of the low use of ride hailing among the public and the limited availability of testing permits.
The question is whether Europe can catch up, as robotaxi volume deployment is probably five years away. I think European companies will be better positioned in Europe after 2025 but are unlikely to get traction in the U.S., China, Japan, or South Korea. Europe is also behind in robotrucks, but its strong global truck manufacturers, Daimler, VW-Traton, and Volvo, have good relationships with leading software driver platform companies.
This means the European truck manufacturers should be strong competitors in Europe and may also do well in the U.S. and China. A key is to allow more robotruck testing with safety drivers in Europe. Europe should be a leader in fixed-/flexible-route AVs, as these are a great match for the European mass-transit industry.
However, the AV shuttles developed by EasyMile and Navya are currently too expensive for volume use. Will any major European automotive company step up to kickstart the volume stage? Some OEMs are likely to provide van-based AVs for fixed-route AV applications; VW may head that list.
Tier 1s such as Bosch, Continental, and ZF may be on the list as well.
California AV test results
Cruise: More Robotaxi Data
Europe's draft regulation
OEM activities with AV software driver developers
Egil Juliussen is the former director of research for infotainment and ADAS at IHS Automotive; an independent auto industry analyst; and EE Times' "Egil's Eye" columnist.