Sunday, February 23, 2020

The Race Towards Autonomy

A promising start but the finish line still some miles away


By Martin Tonko


There was a time when automakers worldwide were pushing the boundaries of how quickly they could realize autonomous driving technology. Various prototypes were developed and tested. Will autonomous cars have a steering wheel? Will the driver and front passenger seats resemble an airplane cockpit or a living room with a giant screen? But as the world moves toward the autonomous era at a greater velocity, innovators are starting to realize the need for a more measured approach.


Initial innovation in autonomous driving is well underway, although the technology’s safety still needs to be assessed. Companies in the industry are more conscientious than ever, thinking seriously about what is technologically feasible and socially appropriate in the coming years.


As the industry pursues the driverless car dream, it is worthwhile to unpack the work that goes on under the hood.


Automotive disruption is not just autonomous


Autonomous driving will profoundly transform global transportation along with three other waves of disruptive technologies, namely mobility, connected cars, and electrified vehicles. These four key industry trends are often abbreviated as MADE. While companies have achieved viable solutions for mobility, connectivity, and electrification, autonomous driving still remains in its early stages.


Mobility: The sector is maturing


The world has seen substantial progress in mobility over the past few years. The ride-hailing fleets of companies such as Uber, Didi-Chuxing, GO-JEK, and Grab are dominating the roads.


As the market matures, some form of regional consolidation is starting to take shape. In China, early innovator Uber relinquished its foothold to local competitor Didi-Chuxing. A similar scenario occurred in SEA, where regional player Grab gained majority control of the market at Uber’s expense.


In terms of investment, many agree that venture financing in mobility tech has moved past its peak.


Connectivity: Well beyond the pilot stage


Like future cities, future vehicles are poised to be smart and connected. They will be equipped with software-based digital solutions, ranging from GPS tracking and speed control to accident recognition and vehicle diagnosis.


Today, connected cars already represent more than half of new vehicles sold globally; by 2025, virtually every new vehicle sold is anticipated to fit this description (Roland Berger). Connected vehicles are already impacting the automotive supply chain, due in part to high consumer demand.


Electrified: Growing across all categories


The automotive industry has embraced vehicle electrification to some degree. As it becomes mainstream, no successful automotive original equipment manufacturer (OEM) brand can afford to not explore, launch, and market a fully electrified vehicle. Maturing battery technologies, falling manufacturing costs, rising fuel prices, and consumer demand have all made electrified vehicles a compelling option that can’t be ignored.


Autonomy: A promising start


To date, the industry has only achieved a limited form of autonomy in very defined domains and no dominant system design for autonomous vehicles has been found. VC investments in complementary AI technologies are still growing and have yet to peak, meaning there are vast opportunities ahead.


In terms of governance, no government has regulation in place beyond vehicle testing. On the safety front, autonomous vehicle insurance products haven’t been developed. Commercialization is unlikely in the foreseeable future, and the overwhelming consensus is that the industry is still in its infancy.


Early days for driverless technology


While the race for autonomous technology is very much alive, will the world see a breakthrough soon?


Of the many different real-world scenarios to be tested, the industry is paying close attention to one critical and extreme use case: urban city roads with pedestrians. A breakthrough will only arrive when the technology is reliable and safe enough to be licensed for operation in these complex traffic situations. Once the first license is extended, large investors will likely rush in, bringing the technology into markets across the globe.


Today, the technology is still confined to testbeds in defined ranges and road conditions, and it must advance further to do away with safety drivers. For autonomous vehicles to scale, a specific suite of technologies must also improve, one of which is computing power.


Intelligent autonomous cars need to sense and adapt to real-time driving conditions, requiring a computing speed of at least 10.0 gigahertz (GHz). As a reference, the average processor speed for desktops is currently 1.5 to 2.5 GHz and 1.0 GHz for laptops (Lehigh University). Other interdependent technology enablers include sensor miniaturization, signal output quality upgrading, and multi-sensor signal processing. They must be enhanced to limit the garbage-in garbage-out effects of information processing.


Algorithmically, autonomous vehicles will eventually be infused with AI systems, which provide the underlying technologies to link system reasoning to regulation, namely pattern recognition, semantic segmentation, and deep learning. As such, existing limits of AI affect what is possible for autonomous driving. Once this bundle of technologies improves and becomes explainable, it will provide the basis for legal discussions in court cases.


Autonomous cars need to reliably function using standalone vehicle-mounted sensors as exclusive data sources, anytime and anywhere. 5G can enhance performance, such as enabling constant vehicle-to-vehicle and vehicle-to-infrastructure signal communication for increased stability and convenience.


However, fully autonomous vehicles need to show that they can be independent and not reliant on 5G or other systems that they cannot control. In light of this fact, market-readiness is not expected to happen before 2025.


A tale of accountability, financing, and governance


Suppliers of the automotive supply chain are accountable to the probability of their product going rogue. When technology fails, suppliers have to cope with the fallout of vendor recalls, which often result in high insurance payouts. Autonomous vehicle suppliers will be no exception. Consequently, they will need to set aside a significant amount of capital to prepare for the liability posed by component failures.


The rise of autonomous vehicles is closely intertwined with AI system development, which is still primarily dependent on VC financing into startups. AI needs to migrate to large enterprise players with deep capital reserves, be it through mergers or acquisitions. Large automotive OEMs, mega suppliers, and tech giants have the scale to cope with the obligations of a vehicle launch. Since startup funding is still on the rise globally, market consolidation around key major enterprises will likely happen in the more distant future.


Licenses will only be granted if vehicle compliance with regulation can be proven. In practice, only a few nations have tinkered with regulating autonomous driving technology. The U.S. Department of Transportation has started a dialogue for ‘Autonomous Vehicle 3.0’ guidelines that will clarify R&D parameters. China’s Ministry of Industry and Information Technology has similarly published ‘National Standards for Autonomous Vehicle Testing.’ Singapore has issued ‘Technical Reference 68,’ which will inform the regulation of autonomous vehicles at some point in time.


That said, vehicle testing is still the focus and many more assessments will be needed before governments can arrive at a comprehensive regulatory framework for autonomous products.


Out with the old, in with the new?


The autonomous era is already in motion and it is worthwhile to look beyond 2025. How will the actors in the industry respond when autonomous vehicles arrive?


The potential is enormous. Once a dominant autonomous vehicle design is found, the success will drive rapid commoditization of the technology. At that stage, cautious players who are late to the game or have limited R&D budgets will look to procure autonomous driving functionality as a system, mainly via turnkey solutions from large suppliers or other owners of the technology.


Overall, the autonomous vehicle industry will likely see value creation happening on the perimeter, meaning that business opportunities will arise around passenger convenience, auxiliary infotainment, or efficiency of travel. For instance, tech giants such as Google and Baidu have the long-term inclination to control and monetize the environment of a passenger in transit. An autonomous car will be characteristically similar to a smartphone or a smart TV, becoming a gadget to engage consumers during a significant portion of their waking hours.


How else will autonomous cars transform roads and businesses? Who is accountable for the safety of autonomous vehicles? These are real questions that need to be answered as the industry moves beyond its early stages. This race toward autonomy will be consequential; cars of the future will no longer look and feel the same. Until then, one can only marvel.


About the Author:


Martin is Head of Automotive SEA and a Partner at global strategy consultancy Roland Berger. He focuses on Automotive and Financial Services. Martin has a degree in Computer Science from Karlsruhe University in Germany, and a Ph.D. in Engineering. Martin is working out of Singapore and Jakarta.

Email This Post Email This Post

Review overview

Sorry, the comment form is closed at this time.