AV Platform An Overview of Autonomous Car Tech Platforms—EMEA, Part I
Plenty of automakers have announced or shown off self-driving vehicles. But the majority of these companies are not creating their own self-driving hardware and software systems in-house; they’ve either acquired specialized firms to create these systems for them or are outsourcing the work to third parties. Some of the most respected leaders in this growing market segment are in the Europe, Middle East, and Africa (EMEA) region.
There’s lots of talk about autonomous vehicles (AVs) these days, and a fair number of prototype cars and trucks are currently on public roads. There are also fully functioning autonomous robo-taxi services ferrying passengers around various global locales, occasionally without anyone in the driver’s seat.
But for all the talk and test vehicles in the marketplace, the number of actual “platforms”—the intelligent technology at the heart of AVs—remains limited. The reason for this is simple—creating these platforms is a major endeavor requiring significant capitalization, copious systems engineering, and rigorous testing. For now, this work is better suited to technology companies (ones with experience in computer hardware and software), rather than automakers. (For the purposes of these articles, automakers may be referred to as original equipment manufacturers, or OEMs.)
This explains why some of the biggest car companies in the world have either outsourced their tech platforms or have purchased companies specializing in creating them.
A new awareness of the limitations of the technology has led a number of vendors to concentrate their efforts on delivering a fleet of robo-taxi AVs rather than production vehicles in order to gain valuable real-world testing experience. Another reason for this focus on fleets and taxi services is that in the future, individual mobility may decrease in favor of multi-person or group mobility.
Some companies have chosen to collaborate or share data with their former competitors in an effort to lower the costs of developing their own AV platforms. This has created an often confusing web of alliances, licensing agreements, and investment partnerships in the automotive industry.
This article is the third in a series comparing and contrasting the dominant industry players and their technology platforms, often referred to in the industry as “stacks.” It explores companies based in the Europe/Middle East/Africa (EMEA) region, while additional articles in this series examine those in North America and in Asia. Specifically, this article covers AImotive, Aptiv, Bosch, Continental, Daimler/Torc Robotics, FEV, and FiveAI.
EMEA AV Tech Platform Players, Part I
The following companies have developed a broad range of AV platform technology for disparate OEM purposes. For some firms, production cars for consumers are the primary target application. For others, the main focus may be autonomous trucking, suburban ridesharing, middle-mile delivery, or low-speed shuttles.
Note that not every company listed below produces or incorporates a complete standalone platform. But each produces critical hardware and software that enable AV platform functionality.
Headquartered in Budapest, Hungary, AImotive was spun off from computer benchmarking firm Kishonti (named after AImotive’s CEO Laszlo Kishonti) in 2015. Investors in AImotive include Bosch (see below), NVIDIA (see the second North America article in this series), and the European Investment Bank.
Prior to its spinoff, in 2014, AImotive created a prototype Mercedes-based AV that it tested on Hungaroring, the Formula 1 Grand Prix track in Hungary.
In 2016, AImotive announced it was working in conjunction with car safety system vendor Autoliv and NVIDIA to integrate AImotive’s artificial intelligence (AI) software into XC90 model SUVs from Volvo (see the Asia articles in this series as Volvo is now Chinese-owned). This announcement was part of Volvo’s 100-driver Drive Me pilot AV program.
Drive Me is a program to put SAE Level 2 SUVs into the hands of 100 “average” Swedish, Chinese and British family drivers. Initially, the program was going to utilize SAE Level 4-capable vehicles, but this idea was eventually scaled back.
aiDrive, aiSim, and aiWare
After development work of nearly a decade, in December 2017, AImotive rebranded its scalable, hardware-agnostic SAE Level 4-capable AV software platform aiDrive.
aiDrive processes sensor data in real-time to provide 360-degree detection of vehicles, lanes, pedestrians, signs, and more, as well as enabling autopilot functions.
Besides aiDrive, AImotive’s purpose-built, non-gaming virtual simulation environment is called aiSim.
AImotive’s flexible, licensable, ASIL-D-capable accelerated neural network AI inference engine is called aiWare. aiWare is highly efficient, optimized for production AVs, and uses a Khronos Group NNEF-based software development kit (SDK). Test chips for aiWare have been developed in conjunction with chipmakers VeriSilicon and GlobalFoundries.
“AImotive has built a range of scalable self-driving solutions based on our highly modular and comprehensive software toolkits. Our customers are able to mix and match their capabilities with ours to create complete, differentiated embedded solutions,” explains Peter Kovács, AImotive’s Chief Product Officer.
In 2017, AImotive partnered with France’s PSA Groupe (owner of the Peugeot, Citroën, Opel, and Vauxhall brands) to develop an AI-enhanced SAE Level 4-capable highway autopilot prototype based on a Citroën C4 Picasso vehicle that could drive at speeds up to 80 mph/128 kph.
AImotive supplied the software AV platform and sensor hardware for the PSA Groupe project (PSA Groupe is also working with Aptiv [see below] on AVs).
At the 2018 Consumer Electronics Show (CES) in Las Vegas, AImotive announced it would provide software for South Korean electronics conglomerate Samsung/Harman’s DRVLINE AV platform.
As of 2018, AImotive was using Toyota, Citroen, and Volvo cars to test its products with. The company has gained the right to test AVs in Hungary, Finland, and the U.S. states of Nevada and California. The company has offices in these locations and in Tokyo, Japan.
In May 2019, AImotive announced it would be utilizing the high-performance V32 automotive-grade processing hardware from Taiwan’s Quanta Computer for aiDrive. Quanta, in turn, uses components from Intel and NVIDIA (see the North American articles in this series) in its system solutions. (It’s rumored that Quanta hardware is also being used for AV development at Apple in the United States.)
In September 2019, AImotive announced it would work with Swedish sensor and positioning vendor Hexagon and French-Italian semiconductor firm STMicroelectronics on a localization and positioning engine for AV platforms for challenging environments such as cities and parking garages.
In October 2019, AImotive announced it would partner with ON Semiconductor on AI-based AV sensor fusion. South Korean fabless semiconductor firm Nextchip announced it would be integrating aiWare accelerator technology into Nextchip’s SAE Levels 2 to 4 Apache5 Imaging Edge Processor system-on-a-chip (SoC) for carmaker OEMs and Tier 1 suppliers.
In November 2019, AImotive demonstrated a proof of concept of a robust, scalable SAE Level 4 automated valet service, developed in conjunction with telecom Vodafone and Dutch mapping concern HERE (HERE is co-owned by Audi, BMW, Continental [see below] and Daimler [see below]).
The idea is that AImotive’s aiDrive AV platform would utilize parking infrastructure maps and free parking space information supplied by HERE and delivered by narrowband Internet-of-Things (IoT) communication by Vodafone. The technology could one day be licensed to automaker OEMs and/or Tier 1 suppliers.
This former parts division of General Motors purchased Carnegie Mellon AV spinoff company Ottomatika in 2015. In late 2017, Aptiv acquired Massachusetts Institute of Technology (MIT)-associated AV startup NuTonomy.
Aptiv now has a deal in place with Hyundai to produce a proprietary robo-taxi-specific AV platform by 2022. In the meantime, ridesharing firm Lyft presently operates a fleet of BMW Series 5 robo-taxis in Las Vegas, Nevada that makes use of Aptiv’s current Centralized Sensing Localization Planning (CSLP) AV platform.
Like Waymo’s platform (see the second North America article in this series), Aptiv’s CSLP has achieved full Society of Automotive Engineers (SAE) Autonomy Level 4+ (“fully automated driving”) functionality in a relatively small geofenced operating area in Las Vegas. For now, Aptiv keeps humans behind the steering wheel to serve as backup drivers in case of system failures.
Venerable German tool and parts maker Bosch has been producing and selling AV hardware, including sensors and ADAS, for many years now. By 2019, Bosch estimated that its sales of ADAS had reached €2 billion annually. Many of Bosch’s ADAS components are designed with safety redundancy in mind, so vehicles they’re installed in can still escape hazardous circumstances when human operators are unable to retake manual control.
In 2015, former Bosch board member Wolf-Henning Scheider (who is now Chairman and CEO of AV platform company ZF Friedrichshafen—see the second EMEA article in this series) stated that Bosch would deliver SAE Level 5 driverless AV technology to the automotive marketplace by 2025. This timeline was later revised to 2030, but it’s clear that Bosch has been making significant strides in this direction in the meantime.
As of early 2013, Bosch was the first automotive supplier to test AVs on public roads in the U.S. and Germany. In 2015 and 2016, Bosch invested in AV platform company AImotive (see above).
Collaboration with Daimler
Currently, Bosch and Daimler (see below) are collaborating on an automated valet parking service for parking garages, with a first system scheduled for operation at the Mercedes-Benz Museum in Stuttgart. Both companies presently share office space in Stuttgart and in Sunnyvale in Silicon Valley and until early 2019 were co-owners of the German electric motor manufacturer EM-motive (Bosch assumed full control in late January).
Since 2017, Bosch and Daimler have been working together with American AV platform maker NVIDIA on the latter company’s Pegasus AV platform. Originally, the platform was going to support SAE Autonomy Levels 4 and 5 in production vehicles for individual consumers. However, over time, both companies realized that this goal was unrealistic, at least in the short term.
Instead, for now, Daimler and Bosch are establishing a Level 4 app-ordered ride-hailing service in San Jose, California. This service will make use of AV lessons learned at Daimler’s 100,000-square-meter testing ground in Immendingen, Germany.
Eventually, it’s thought that as the ride-hailing service proves itself in California, Daimler can take the proprietary platform and roll it into a package aimed at third-party taxi fleet operators. In turn, it’s hoped that these operators would buy self-driving vehicles from Daimler. (Daimler also currently distributes a mobility-as-a-service [MaaS] app for non-AVs in San Francisco and Berlin.)
In the meantime, Bosch is also working with BMW on the latter’s line of AVs called iNEXT, which are scheduled to come to market—starting with an SAE Level 3 car—in 2021.
Bosch in Silicon Valley
Currently, Bosch is a major contributor to the Stanford Center for Automotive Research in California as well as to the Silicon Valley Robotics Forum (SVRF) and the IEEE Robotics and Automation Society.
In August 2018, Bosch invested in DeepMap, a Silicon Valley startup that provides mapping and localization services for AVs. DeepMap’s technology creates continuously updating HD, centimeter-precise maps in real-time aboard a vehicle and in the cloud. A number of DeepMap’s executives previously worked in key positions at the mapping divisions of AV-related companies Google, Apple, and Baidu (see the first Asia article in this series).
Bosch currently sells or has in development electronic control unit (ECU), V2X, and intelligent cockpit components for AVs.
At the 2019 CES, Bosch presented an electric driverless mobility shuttle AV concept that could connect via V2X communications to local surroundings. Bosch stated that it desires to provide the vehicles, tools, and services necessary to enable AV ride-sharing services in the near future.
In late 2019, Bosch announced it was working on driver and passenger monitoring and alertness systems to be introduced prior to EU laws mandating them for European vehicles by 2022.
Relationship with Tesla
Bosch has had somewhat of a contentious relationship with AV manufacturer Tesla (see the second North America article in this series).
In 2015, Bosch took two Tesla Model S and added 15 sensors and LiDAR to each one to create Bosch prototype AVs.
In March 2018, Tesla blamed rusty Bosch parts for a recall of some 125,000 of its Model S vehicles.
In 2020, Bosch CEO Volkmar Denner said that a new Tesla factory near Berlin could be a new sales opportunity for Bosch as the auto industry gravitates toward electric vehicles.
German car components and technology company Continental is a market leader in ADAS, which it has been working on for more than two decades. Continental demonstrated driverless AVs as long ago as 1968. The company tested AVs in the United States as early as 2012 and in Japan as early as 2014. Continental is firmly committed to the AV industry Vision Zero goal of zero accidents caused by autonomous technology.
The company is also a big believer in AI, and it conducts joint research activities with AV platform makers NVIDIA (see the second North America article in this series), Baidu (see the first Asia article in this series), Oxford University, the DeepDrive Center at the University of California at Berkeley, the Technical University of Darmstadt, the Indian Institute of Technology at Madras, and the German Research Center for Artificial Intelligence (DFKI). Continental has pledged to have at least 700 in-house AI specialists on staff by 2021.
In 2016 and 2017, Continental initiated the online 2025AD discussion platforms for interaction between developers, academic researchers, and end-users of AV technologies. Located at 2025AD.com and 2025AD.cn (the latter in Chinese) and accessed by more than 10,000 users per month, they’re leading resources for AV information.
ADCU and SDCU
Continental’s primary AV platform is called the Assisted & Automated Driving Control Unit (ADCU). Co-developed by Silicon Valley logic device supplier Xilinx, the highly customizable, scalable and programmable ADCU makes use of the Khronos Group-standardized Open Computing Language (OpenCL) framework. The ADCU is imbued with extensive AI, neural networks, and machine learning provided by NVIDIA DRIVE SoC and software components.
The ADCU is also designed to be ASIL-D safety-compliant; with the latest Continental sensors, it can discern intentions of pedestrians based on their movements and gestures. There’s also driver monitoring via a vehicular interior camera. The platform uses technology from Continental’s Argus Cyber Security subsidiary for safety from hacking and external manipulation.
In addition to the ADCU, Continental has developed a crash-proof Safety Domain Control Unit (SDCU) as a fallback safety mechanism in the event of an ADCU failure. It’s also designed to function if a vehicle’s driver is unable to take over manual control of a vehicle for any reason. This level of redundancy and system diversity has been extensively proven in aviation systems.
Continental has been testing the ADCU and SDCU with SAE Levels 3 and 4 automation in its manned Cruising Chauffeur vehicles on both highways and urban streets for the last several years. In 2017, the company demonstrated cloud-driven automated valet parking functionality and safety electronic control units (ECUs) via its wholly owned AV technology subsidiary Elektrobit.
Continental is aiming to introduce an SAE Level 3-capable ADCU/SDCU-equipped AV platform based on the Cruising Chauffeur in 2021.
BMW and Intel/MobilEye Partnership
In June 2017, Continental joined BMW and Intel/MobilEye (see the first North America article in this series) as a development partner for a scalable AV solution platform that originally was announced in 2016 for BMW AVs launching in 2021. For this platform, Continental was seen as a system integrator providing function development, motion control, simulation, and validation. Continental and BMW had already worked together on a number of other projects and research. However, it’s unknown what the current status of this collaboration is; since this announcement, BMW has made various agreements with other AV platform vendors and carmaker OEMs.
In July 2017, Continental made an undisclosed investment in Toulouse, France-based AV platform company EasyMile (see below). EasyMile is the maker of 15-passenger, electric driverless EZ10 shuttle AVs that have been deployed in more than 50 cities worldwide. The two companies signed a memorandum of understanding in 2018 to collaborate on driverless AVs and mobility solutions at a dedicated research and development Center of Excellence (CoE) office in Singapore. There and elsewhere, their mobility solutions will be rolled out for public and private use. Continental does AV testing and maintains other AV Centers of Excellence in Hungary, Germany, China, Japan, and the United States.
At the 2018 CES, Continental introduced (via a virtual demonstration) the handicapped-accessible, app-summoned Balanced Economy and Ecology (BEE) 2-passenger electric concept AV microcar for individual door-to-door mobility. Capable of speeds up to 60 kph and equipped with window-as-display infotainment, the BEE is designed for autonomous urban swarming/flocking and can travel up to 350 kilometers daily.
The BEE features Intelligent Intersection technology that integrates powerful sensors, sensor fusion algorithms, and dedicated short-range communication (DSRC) units for V2X broadcasting. Continental announced plans to begin testing its Intelligent Intersection technology in the near future in Columbus, Ohio, a winner of the 2016 U.S. Department of Transportation Smart City Challenge.
NAV and AAI Alliances
In June 2018, Continental formed the Networking for Autonomous Vehicles (NAV) alliance with Bosch (see above), NVIDIA, Aquantia, and Volkswagen to drive development of vehicular Multi-Gig Ethernet networking.
At the 2019 CES, Continental announced a collaboration with Berlin-based startup Automotive Artificial Intelligence (AAI) to work on open, scalable, and modular simulation testing tools for AVs. Whereas AAI is responsible for integrating a virtual environment and simulated traffic for these tools using machine learning, Continental is responsible for AV simulation, including ADAS and driving dynamics.
Continental has a vision of modular driverless AVs using what it calls the Continental Urban Mobility Experience (CUbE) for app-driven passenger shuttles and delivery services that would travel the bulk of a commuter or delivery journey.
For shuttle or robo-taxi services, intelligent displays and audio in the shuttle vehicle would alert passengers with personalized messaging to travel information, notifications, and local services. Even external pedestrians would be acknowledged by the CUbE vehicles using outward-facing displays.
Continental demonstrated EasyMile-manufactured EZ10 electric CUbE shuttle prototypes at the 2019 International Motor Show in Frankfurt, Germany and at CES in Las Vegas.
For the so-called “last mile” of deliveries made with cargo versions of CUbE AVs, the vehicles would offload four-legged robot delivery “dogs” that would take packages to recipients’ doorsteps.
Says Ralph Lauxmann, the head of Continental’s systems and technology for its chassis and safety division, “Our vision of cascaded robot delivery leverages a driverless vehicle to carry delivery robots, creating an efficient transport team.”
“Both [the AVs and the dogs] are autonomous and, in principle, both can be based on the same scalable technology portfolio. These synergies create an exciting potential for holistic delivery concepts using similar solutions for different platforms. Beyond this technology foundation, it’s reasonable to expect a whole value chain to develop in this area.”
Further 2019 Developments
In 2019, Continental acquired a minority stake in Israeli AI object detection/machine learning accelerator startup Cartica AI.
In July 2019, Continental demonstrated AV truck “platooning” functionality in Frankfurt, Germany in cooperation with Munich-based braking systems company Knorr-Bremse. Continental’s ADCU and Knorr-Bremse’s global scalable brake control (GSBC) and driving stability motion control all contribute to platooning capabilities. Platooning allows trucks in a single-file convoy to be spaced closer together, saving fuel and lowering emissions. Platooning will be a part of Highway Pilot — an AV technology for trucks being developed by both firms for demonstration in 2020.
In August 2019, Continental launched a six-month pilot driverless EZ10 CUbE shuttle AV shuttle deployment on the campus of Oakland University in Auburn Hills, Michigan, funded by the U.S. State of Michigan’s PlanetM mobility initiative. The pilot additionally integrated Continental’s EVIR and Zonar Z Pass technologies for inspection/maintenance and passenger entry/exit detection, respectively.
Daimler Trucks/Torc Robotics
Technically based in Blacksburg, Virginia in the United States, Torc Robotics is majority-owned by the Trucks Autonomous Technology Group of Germany’s Daimler (inventor of the truck).
In 2014, Daimler presented its prototype Mercedes-Benz Future Truck 2025—the world’s first automated truck.
In 2015, Daimler demonstrated the Freightliner Inspiration Truck, which obtained the first-ever road license for a partially automated commercial vehicle. It also demonstrated the Mercedes-Benz Actros heavy-duty truck with Highway Pilot assistance and connectivity systems on public roads.
With its Active Drive Assist system on Actros and FUSO Super Great trucks and its Detroit Assurance 5.0 with Active Lane Assist on its Freightliner Cascadia models, Daimler is the first vehicle manufacturer to bring SAE Level 2 capabilities to production trucks. These systems can accelerate, brake and steer autonomously in all speed ranges.
Founded independently in 2005, Daimler’s Torc Robotics has in the past provided more than 100 autonomous solutions for the mining and defense industries. These systems closely focused on safety and effectiveness. Torc has also worked with construction vehicle company Caterpillar for more than a decade.
In 2007, Torc partnered with Virginia Tech University to enter the DARPA Urban Challenge 60-mile AV race, finishing third in 4 hours and 36 minutes. In 2011, the two organizations partnered again to enable a blind driver to successfully navigate the Daytona International Speedway for the first time.
Torc’s demonstration SAE Level 4-capable Lexus RX and Chrysler Pacifica vehicles, which use the company’s Asimov software AV platform, test goal-specific use cases and edge cases for self-driving.
Torc has claimed that Asimov boasts 39 discrete AV functionalities as well as localization, mapping, real-time navigation, and obstacle detection and tracking (including vehicles driving in the wrong direction).
“While a lot of startups may have interesting pieces of technology, we have a decade of experience delivering an end-to-end [AV] solution that actually works. This includes multimodal sensor fusion (vision, radar, and LiDAR) and the use of both traditional autonomy algorithms, coupled with newer machine-learning techniques,” says Torc Chief Technology Officer Ben Hastings.
Torc has tested its AVs in 20 U.S. states, both on highways and in urban environments, and in inclement weather, including snow and heavy rain. It also has conducted extensive testing in virtual environments.
In 2017, Torc signed an agreement with semiconductor firm NXP (see the second EMEA article in this series) to use NXP’s radar with Torc’s Asimov platform.
In the 2018 CES in Las Vegas, Torc partnered with AAA Northern California, Nevada & Utah (NCNU) to develop safety criteria for AVs via testing of vehicles in Las Vegas, Nevada and other cities.
At the 2019 CES, Torc announced a partnership with global transport and mobility operator Transdev to deploy SAE Level 4 fully autonomous 16-passenger i-Cristal shuttle vehicles in Paris and Rouen in France.
Lacking steering wheels and foot pedals, the AVs were co-developed by Transdev and French vehicle manufacturer Lohr, using the Torc Asimov AV platform. They will travel in mixed traffic in Rouen and in dedicated lanes in Paris at speeds up to 19 mph/30 kph.
Transdev, which provides 11 million passenger trips daily in 20 countries (some of which take place in AVs), has stated that if successful, it would like to additionally deploy i-Cristal shuttle fleets on many of its other global routes and transport networks.
Another development partnership for mobility was announced at the 2019 CES between Torc and Chinese electric vehicle (EV) manufacturer Bordrin. Initially, vehicles developed will feature SAE Level 2+ functions, but will eventually offer full SAE Level 4 functionality. They will also make use of intelligent transportation systems (ITS) for smart cities.
In the Fall of 2019, Torc tested SAE Level 4-intended trucks on public roads in southwest Virginia, following extensive previous testing on a closed-loop track. The trucks contained both a safety driver and a system engineer. The tests also utilized a main control center and several logistics hubs along high-density truck freight corridors, which would likely continue to be used in future autonomous truck operations.
Founded in 2014, electric autonomous shuttle company EasyMile SAS is based in Toulouse, France. The company’s primary 15-passenger EZ10 driverless AV was developed out of the CityMobil2 project co-funded by the European Union’s Seventh Framework Programme for research and technological development.
The EZ10 uses camera, odometer, inertial measurement unit (IMU), GPS, and LiDAR sensors as well as 4G V2X communications to stay in touch with its environment and supervision center.
Road obstacle detection forces the vehicle to either adapt its trajectory, change its speed, or apply its brakes. Redundant and fail-safe systems maximize safety for occupants and other road users. The EZ10’s AV platform includes embedded cybersecurity.
The position, usage, and assignments of each EZ10 can be closely monitored; maximum speed is 25 kph. All EZ10s are equipped with a “Black Box” for recording data in the event of an accident.
In 2015, the Gelderland county of the Netherlands announced it would deploy the EZ10 on a 7-km route between the campus of Wageningen University and Ede-Wageningen railway station. A second pilot program used two EZ10s on a short route in the Dutch village of Appelscha.
In 2016, EasyMile’s first North American deployment was in the Bishop Ranch Office Park in San Ramon, California. In 2017, the EZ10 was deployed in Tallinn, Estonia, where it ran in mixed traffic, and on the campus of National Taiwan University in Taipei.
In all, the EZ10 has had more than 200 deployments and has driven more than 600,000 kilometers autonomously. It currently runs in 25 countries globally.
Investments and TractEasy
In January 2017, French transport multinational firm Alstom invested in EasyMile, and the companies signed a partnership agreement. Later that year, German AV platform maker Continental (see above) invested in EasyMile. Continental has collaborated with EasyMile on AV technology and has had its own deployments of EZ10s.
Later in 2017, EasyMile launched the 25-ton traction capacity TractEasy electric tow tractor AV. The TractEasy is an autonomous version of the market-leading Jet 16 EV by French industrial equipment manufacturer TLD Group. Like the EZ10, the TractEasy has a full complement of sensors and is designed for indoor and outdoor operation at speeds up to 25 kph in mixed traffic in a limited area. Like the EZ10, the TractEasy can be managed by a central supervision center.
Future driverless projects for EasyMile will feature trams and buses.
German engine and automotive component development firm FEV has been working on ADAS and AV technology since 2016. FEV has designed sensors, algorithms, redundant electronic architectures, complex deep-learning systems, and 5G vehicle-to-infrastructure (V2I) connectivity solutions for AVs (which it refers to as “smart vehicles”).
A Unique Approach
Unlike other automotive component companies, in the past, FEV has traditionally been retained as a service provider on a project basis by carmaker OEMs and Tier 1 suppliers, rather than manufacturing parts to sell to them.
Unlike other AV platform developers, FEV sees the possibility (which it recommends) for its customers to “jump” directly to SAE Level 5 (“fully driverless”) AVs, rather than evolving them gradually by developing vehicles capable of SAE Levels 1 to 4 first. Indeed, since SAE Level 5 AVs don’t need a driver at all, the option of relinquishing such duties to a human no longer applies.
In many ways, FEV believes this only simplifies and streamlines the development of an AV platform. Recognizing that this method has the potential to bring with it increased risks to safety, FEV places a special focus on extensive validation and testing.
The company uses hardware-in-the-loop solutions, simulation and modeling tools, Microsoft Azure cloud-based data logging, and automation to enable this.
Testing and SVEN
In 2017, FEV built a small AV fleet that it tested on roads in the U.S.A., Turkey, and Poland. These cybersecure, SAE Level 5-capable and vehicle-to-everything (V2X)-connected cars were capable of recognizing traffic signs, objects, and people.
Appropriate proactive and reactive maneuvers were performed. Rule-based and AI/machine-learning approaches for control algorithms were compared and contrasted.
At the end of 2017, FEV purchased a majority interest in Aachen University of Applied Sciences mobility spinoff Share2Drive. In 2018, Share2Drive announced the three-passenger, networked, SAE Level 4-ready Shared Vehicle Electric Native (SVEN) EV microcar. Fully customizable by FEV for large customers, the SVEN includes four cameras, one mid-range radar, and seven LiDAR sensors. In 2019, FEV announced it would sell its stake in Share2Drive, although the company would continue plans to begin marketing SVEN in the third quarter of 2021.
Later in 2018, FEV opened an AV testing Center of Excellence (CoE) in Auburn Hills, Michigan as an addition to its systems engineering office in California’s Silicon Valley.
In 2019, FEV became an integrator of PolySync Technology’s DriveKit AV testing system.
At the 2020 CES, FEV showed driver monitoring and alerting systems, proprietary V2V technology, a CADET gateway module, and a Microsoft Azure-driven AV functionality validation system.
AV software startup FiveAI was founded in 2015, in Cambridge, England by CEO Stan Boland, who sold his previous company, Icera (a fabless semiconductor developer) to NVIDIA (see the second North America article in this series) in 2011.
No Maps Required
Originally, FiveAI’s original approach to autonomous operation was unique due to the lack of a need for extensive 3D environmental mapping; instead, its platform generated maps in real-time on-the-fly. The company’s AV platform makes use of input from LiDAR, cameras, radar, GPS, inertial measurement units (IMUs), and vehicle telemetry. It utilizes high-density computation, strong AI, and lab-refined research to achieve safe, reliable, low-disengagement mobility.
Back at FiveAI’s AV garages, real-world testing data is filtered, prioritized and compressed before being uploaded to an AWS cloud via Kubernetes-powered edge computation. Parallel-operating machine-learning models are trained using Horovod and Spark open-source frameworks and dataframes.
Further system trials at FiveAI are conducted using simulations, with the company employing multilayer deep-generative model replicas of London streets. Driving scenarios can be tested with a billion different variable permutations (including the addition or subtraction of cyclists, pedestrians, vehicles, roadworks, foliage, and weather elements) to strengthen all-important safety performance.
In 2017, the British government established an AV consortium called SteetWise to test AVs on public roads in the UK. Members of Streetwise include FiveAI; McLaren Applied Technologies, the University of Oxford, Arriva, TRL (formerly Transport Research Laboratory); Transport for London; and Direct Line Group, a car insurer. The goals of Streetwise are to make the UK a world leader in AV research, develop methods to validate AV safety, and create models for AV insurance and service.
Initially, a 19-kilometer route was chosen between the British boroughs of London, Croydon, and Bromley. Starting in August 2018, five SAE Level 3-capable FiveAI cars picked up passengers at predetermined locations along the route. The purpose of the operation was to gather data on road users and environments, and the cars were all manned by trained safety drivers who kept their hands on the vehicles’ steering wheels even when they were operating in autonomous mode.
A Different Outlook
Says FiveAI CEO Stan Boland, “We developed a much more profound understanding of what it’s really going to take to make self-driving cars safe—and it’s one to two levels deeper than what the literature says. Until you start building a system and having to rely on it and provide evidence of safety, you’re not really tackling it. We learned a lot in the process of doing that.”
Boland sees his company’s UK location as an advantage to developing for the UK and European markets. Still, he doesn’t foresee FiveAI as one of a dozen or so global AV “super system integrators” of the future, but rather as a B-to-B solution provider that can serve such companies.
“We’re able to explain how we view the problems [of real-world trials] and how we turn that into software that’s useful for other people,” he says. A significant portion of the company’s focus is on the academic foundations of autonomous driving; 40 percent of FiveAI’s 115 employees hold PhDs.