Marc Andreessen once famously noted: “Google is working on self-driving cars, and they seem to work. People are so bad at driving cars, that computers don’t have to be that good to be much better.” So when can we push a button and be picked up by a driverless vehicle? The answer is, I don’t know. So let’s find out with a deep-dive into the world of mobility services and interview BESTMILE‘s Raphaël Gindrat.
According to Intel’s CEO Brian Krzanich, autonomous vehicles will generate and consume roughly 40 terabytes of data for every eight hours of driving. The averagely driven car will churn out 4,000 GB of data per day, as it is equipped with hundreds of sensors. Cameras alone will generate 20 to 40 megabyte per second, and the radar, as well as sonar, will generate between 10 and 100 kilobytes per second. Thus the worlds of tech companies and traditional automotive manufacturers are colliding to shape the future of mobility, in which autonomous driving technology replaces humans and turns our cars into mobile server rooms.
The flood of data in autonomous vehicles; Source Intel
Before we dig deeper into the driving technology, let’s agree on a common terminology. The Society of Automotive Engineers (SAE) International has defined six levels of automation – from no automation (Level 0) to full automation (Level 5). These base definitions describe categorical distinctions for a step-wise progression through the levels and are consistent with current industry practice.
Six levels of automation; Source: Society of Automotive Engineers International
A quick elaboration on the above graph. Level 1 systems include technologies like passive cruise control systems, while Level 2 includes many of today’s OEMs’ so-called “active” safety functions, like active radar-based cruise control, autonomous braking, pedestrian detection, and lane-keep assist. Tesla’s autopilot falls into Level 3. While it is capable of driving autonomously on certain roads, it still requires a human driver. The system can flow with traffic, stay in its lane, avoid cars in your blind spot, and of course brake to avoid a collision, but it cannot change lanes without a human command. While Tesla’s system is by far the most advanced semi-autonomous car on the road today, Elon Musk stated repeatedly: “Autopilot is what we have in airplanes. For example, we use the same term that is in airplanes where there is still an expectation that there will be a pilot. So the onus is on the pilot to make sure that the autopilot is doing the right thing.”
The key difference between Level 4 and Level 5 autonomous vehicles then is that the former allows — but does not need — a human to take over, while Level 5 vehicles prohibit human intervention. Google’s self-driving technology WAYMO aims for Level 5. According to Google’s website, their technology accumulated the equivalent of over 300 years of human driving experience, largely on city streets. That’s on top of 1 billion simulated miles that Google drove in 2016 alone. In other words, each self-driving car that is equipped with Google’s technology will have hundreds of years of driving experience the moment it starts operating. And the learnings of all vehicles will be shared across the entire fleet, resulting in an exponential growth of driving experiences. So Marc Andreessen might have a valid point after all.
As we see above, this transformation comes in stages, requires significant investments and offers a huge market potential for the emerging ecosystem. Roland Berger forecasts that by 2030, the new opportunities from autonomous driving will amount to USD 40-60 bn. This includes hardware components, such as cameras, sensors, communication systems as well as software that replaces decision making (machine learning) and navigation.
Before we see fully autonomous vehicles (SEA Level 5) on our roads, engineers still have to solve essential technical problems. For example, the safe operation in heavy snow or rain conditions as well as during pitchfork nights is a risk with the current hardware and software technology. The installation of wayside communication equipment might be required. Furthermore, cyber security issues have to be overcome to improve the overall integrity of self-driving systems. Finally, millions of testing miles will have to generate the statistical data needed, to convince regulators and lawmakers about the safe adoption of self-driving vehicles. Even though companies like UBER are accepting penalties to accelerate their testing, it is unlikely to see fully autonomous vehicles (AVs) to be commercially available before 2025. Meanwhile, advanced driver assistance systems (ADAS) will play a crucial role in preparing regulators, consumers, and corporations for the medium-term reality of cars taking over control from drivers. ADAS then provides vehicles with four critical capabilities:
- 1.Vehicle's location and environment
- As there would no longer be active human input for vehicle functions, highly precise and real-time information of a vehicle’s location and its surrounding environment will be required (e.g., road signs, pedestrian traffic, curbs, obstacles, traffic rules).
- 2. Prediction & decision algorithms
- Advanced concepts based on Artificial Neural Networks (unsupervised/deep learning, machine learning) will be needed to create systems to detect, predict and react to the behavior of other road users, including other vehicles, pedestrians, and animals.
- 3. High accuracy, real-time maps
- Detailed and complete maps must be available to provide additional and redundant information for the environmental models that vehicles will use for path and trajectory planning.
- 4. Vehicle driver interface
- A self-adapting interface with a smooth transition of control to/from the driver, mechanisms to keep the driver alert and a flawless ride experience will be instrumental in winning consumer confidence.
Traditional OEMs have started to develop theses capabilities in-house, but also acquire them externally or through partnerships. Which is why incumbent players today simultaneously compete on multiple fronts and cooperate with competitors. New market entrants are positioning themselves to provide high-margin activities along the value chain and suppliers will increase their bargaining power in the years to come.
Advanced driver assistance systems (ADAS) as well as 1st and 2nd tier suppliers; Source: Financial Times
When being asked for the most advanced OEMs, Raphaël Gindrat answers: “German car manufacturers are very advanced and in a pretty good position. For example, Daimler has just announced an agreement together with Bosch, the largest automotive supplier on the planet, laying out the plan to launch a system that is fully automated (SAE-Level 4) and driverless (SAE-Level 5). The Volkswagen Group is developing self-driving capabilities in-house with AUDI Piloted Driving being at the technological forefront of the group. BMW signed an agreement with INTEL to leverage the artificial intelligence capabilities of WATSON.”
Then Raphaël continues: “GM has bought CRUISE AUTOMATION for USD 1 billion to accelerate autonomous vehicle development. Last year CRUISE employed 50 people, today there run at 500 people head count and in two years they want to triple their size to 1500 employees. FIAT CHRYSLER, for instance, has paired up with Google to develop 100 self-driving minivans and is in discussions with about a similar venture. TOYOTA, TATA, MICROSOFT and GOOGLE all invested in UBER. While these are only a few example of the massive scaling that is happening as we speak. So some OEMs will want to develop the capabilities in-house and others will rely on partners. As for today, Google’s main goal is not to produce vehicles themselves, but they want to become a supplier of the automotive industry providing the operation system for autonomous vehicles in competition to BOSCH, IBM, DELPHI, and CONTINENTAL.”
Investments and partnerships in autonomous cars and ride sharing companies; Source: Bloomberg
Of course, traditional OEMs will do everything they can to ensure that tech conglomerates and software start-ups don’t dominate the future of mobility. They see the threat of being reduced to a sole vehicle supplier. Which is why it is fascinating to watch, how the competitive forces are reshaping the automotive industry and to identify the beneficiaries of novel business models and revenue streams for mobility services.
Asked for his vision, Raphaël Gindrat, elaborates: “Autonomous vehicles are going to change the way people travel in cities. Autonomous vehicles will be able to reduce the total numbers of accidents, relief traffic congestion, reduce pollution and free up parking space. This is only true, however, if vehicles are shared like taxis or traditional public transportation e.g. buses, trams, and metros. If vehicles are still owned by consumers and the total number of vehicles is not decreasing, we will not see a lot of benefits, other than that you don’t have to drive yourself anymore. All the other benefits are linked to the integration of autonomous vehicles into the transportation system of a city, where they are deployed and shared depending on the level of demand.”
McKinsey estimates that rideshare and onboard data services could generate an additional USD 1.5 trillion of annual automotive revenue by 2030, adding to the USD 5.2 trillion from traditional car sales and services. Connectivity and later autonomous technology will increasingly allow the car to become a platform for drivers and passengers to use their transit time for personal activities, which could include the use of new forms of media and services. The increasing speed of innovation, especially in software-based systems, will require cars to be upgradable. It is financially attractive for consumers too: It costs an average of USD 8,558 per year to own a car in the U.S., but each vehicle is used just 4 percent of the time. Ridesharing in an autonomous vehicle could ensure that cars are always in use.
Forecast of diversified automotive revenues until 2030; Source: McKinsey & Company
And the global market for mobility services is growing, too. According to Statista, the total market volume can be divided into four service categories: flights, car rentals, trains and buses, and ride sharing. Comparing worldwide figures, it can be acknowledged that most of the revenue is generated in the United States (USD 127.4 billion), followed by Europe with a revenue of USD 94.3 billion and China with USD 54.7 billion in 2016. Until 2021, all markets are forecasted to grow significantly with annual growth rates of 18.4 percent (China), 10.1 percent (Europe) and 7.9 percent (United States). Now if we subtract the revenue share of flights, which is a market segment unlikely to compete with autonomous vehicle services, we arrive at a total global market volume of USD 120 billion forecasted for 2021.
Revenue forecast for global mobility services by market and user; Source: Statista
By introducing fully automated and driverless driving to the urban environment, mobility service providers will want to tackle a share of the growing market described above. Raphaël Gindrat explains: ” Such mobility companies will want to provide two kind of services: The first service is on demand, for which autonomous vehicles will be dispatched as traditional taxis today. In the second case, vehicles will be used on fixed, stable volume routes much like buses or trams today, being fully integrated into the transportation system of the city.” Being asked about the positioning of BestMile, Raphaël continues: “I often use the analogy of an airport. BestMile is not Boeing or Airbus (OEM), providing the vehicles or the auto-pilot software. We are not the airline because the airline (mobility service provider) is our customer. But our technology works synonymously to a control tower. The control tower has to be agnostic. One control tower has to manage all aircraft, operated by humans or flying on auto-pilot together. At BestMile we are not developing a vehicle, we are not developing any piece of hardware or software that is inside the vehicle, but we are fully committed to bringing a cloud platform to market, which will enable a mobility service provider to operate a fleet of autonomous vehicles efficiently.”
The capability to optimize the fleet operation of autonomous vehicles will be a critical competitive advantage for the mobility service providers in the future. In the past years, new players have entered the global markets with UBER, DIDI and LYFT being the most prominent start-ups. The 10 largest companies managed to attract a total funding of USD 22.4 billion. And the intensity of competition is further increasing.
New entrants in mobility services market worldwide; Source: Statista
As the above companies are in direct competition to the existing market players, Raphaël Gindrat elaborates: “At BestMile, we see four types of potential customers. We have transit agencies, for example, Berliner Verkehrsgesellschaft (BVG), Régie autonome des transports Parisiens (RATP), Transport for London (TfL) who plan, operate and manage public transport in inner cities today. We could also sell to novel mobility service providers like UBER, LIFT, DIDI, etc. as some of them will want external partners to further accelerate their expansion plans. The last type of customers is automotive OEMs themselves. Today you can see that for example Daimler (Car2Go, Mytaxi) and BMW (DriveNow) are diversifying to become mobility service providers. So there will be more and more OEMs that will want to undergo a similar transformation and some of them will use our solution.” Raphaël then expands: “The main opportunity for BestMile is scaling and deploying commercial solutions. Companies like NAVYA, LOCAL MOTORS or EASY MILE are already selling their vehicles and we provide the fleet management system.”
NAVYA EZ10 autonomous vehicle powered by BestMile in operation in Sion, Switzerland; Source: NAVYA
So today, commercially operating autonomous vehicles are rather small shuttles that are replacing bus routes and run at speeds of up to 25 km/h in well-defined areas. However they are real, they run and these projects help all market players to learn and advance the technology. For traditional cars, advanced driver assistance systems will accelerate the transition from humanly operated to self-driving vehicles in the next years. In the years 2025 to 2030 will see a convergence of mobility service providers and the deployment of large fleets of autonomous vehicles replacing taxi, bus and trams drivers in cities across the globe.
Being ask to dream about the future, Raphaël Gindrat advances: “There are a lot of interesting innovations out there, and our priority is laser-focused on people transport with autonomous vehicles. However, with our new investor Airbus on board, who has created a drone concept to transport people, why not leverage our platform to other autonomous vehicles in the future.”