Over the holidays, I started looking at several new technology areas — one of which was autonomous vehicles — and drafted a brief note overviewing what I learned.
In addition to this article, we interviewed Framework member Robert H. late last year about his trip to the Frankfurt Auto Show. Robert, whose first career was as an auto company executive, makes some interesting comments about Tesla and GM — two leaders in the field of autonomous vehicles — as well as comments about electric vehicles and the industry in general. Make sure to listen to that series (videos are split into snippets of only a few minutes each) for some information you are unlikely to hear anywhere else.
In addition, I’ve opened up a Framework Forum thread on autonomous vehicles, so feel free to leave and read comments there as well!
Key Takeaways
- Demographic, economic, and sociological factors in the U.S., combined with advances in sensors and machine vision / machine learning technologies, cheap computer processing power, and the ubiquity of high-speed data connections set the stage for a major disruption to Transportation, Tech and Insurance sectors with the introduction of increasingly capable autonomous vehicles.
- Critical drivers for the adoption of autonomous vehicles include network security, bandwidth, and availability as well as the continued development of machine learning technology.
- Risks to the adoption of autonomous vehicles relate less to technological considerations – Level 3 autonomous vehicles already operate safely in developed countries – and more to legal and sociological ones.
- One of the Big Three has announced plans to commercially test Level 4 autonomous vehicles on public roads within the next 12 months. Sales of Level 4 autonomous vehicles will likely be available to retail consumers by the 2020 model year. We believe Level 4 systems’ first introduction to most consumers will be those used for circumscribed public transportation applications, likely within the next 24 months.
Overview
Autonomous vehicles are enabled by both on-board systems – especially LiDAR (Light Detection And Ranging) sensors, machine vision, and machine learning enabled computer systems – and communication with remote server data via wireless links. The Society for Automotive Engineers (SAE) classifies autonomous vehicles into five levels:
- Level 0: No Automation
- Level 1: Driver Assistance. An automated system on the vehicle can perform specific task-based assistance to the human driver (e.g., back-up obstruction braking)
- Level 2: Advanced driver assistance. The system can perform more advanced task-based assistance to the human driver. (e.g., automatic parking, distance-sensing cruise control)
- Level 3: Limited autonomous driving. The system can drive the car on well-mapped roads with an attentive driver who can take control of the vehicle if necessary.
- Level 4: Conditional autonomous driving. The system can drive the car on well-mapped roads in certain conditions and the human driver need not take control for the journey.
- Level 5: Complete autonomous driving. The system can drive the car on all roads in all conditions without human intervention.
From a demographic perspective, demand for autonomous vehicles is supported by ageing Baby Boomers whose mobility and quality of life will be boosted through this technology.
From an economic perspective, studies have found that personal vehicles spend as much as 95% of their useful lives parked, so even working-age adults may find that on-demand hiring of autonomous cars makes more economic sense than personal car ownership. Exacerbating the opportunity cost of holding a personal car, stagnating wage growth coupled with auto price inflation has meant a doubling of real car prices from 1975-2015, so the economics of autonomous on-demand car-for-hire services look especially attractive. The fact that ride-hailing services like Uber and Lyft, and car share services like ZipCar and GM’s Maven have proven popular, offer anecdotal evidence that attitudes toward car ownership and car sharing may have shifted among younger consumers, especially.
Investors and technologists are attracted to autonomous vehicle development because they see future strong demand for sophisticated digital and analog chipsets, novel software solutions, and opportunities for improvements in network equipment and technology. Set in context of the larger trend toward an Internet of Things (IoT) that allows for autonomous control of various real-world equipment and machinery, autonomous vehicles are an obvious application with an attractively large end market. According to PitchBook, nearly $4 billion worth of venture capital was invested across 68 deals last year, an increase in deal volume of over 500% in dollar terms and 200% in deal terms compared to 2016.
Drivers of and Risks to Widespread Adoption
It is obvious that proliferation of Levels 4 and 5 autonomous cars will necessitate improvements to machine-learning algorithms (i.e., computers making choices when presented with novel and ambiguous situations) LiDAR and the like. However, these on-board systems are only part of the puzzle. Autonomous vehicles must engage in real-time two-way communication with online databases and potentially with other networked devices, so network availability, throughput, and, crucially, security must be improved. It is not a stretch to imagine that the widespread acceptance of autonomous vehicles might be delayed if hackers were able to steer autonomous vehicles into concrete barriers when car owners refuse to pay a Bitcoin ransom or if semis simply stalled in the middle of the road when network signal strength was poor.
Overcoming these technical issues will take time and engineering know-how, but a more salient near-term risk to adoption is the legal status of an autonomous vehicle and the question of what party is held liable in case of an accident. If two autonomous vehicles collide due to a computational or network error, which car owner’s insurance company pays the bill? If an autonomous car wrecks itself and kills its driver to avoid a fatal accident with a group of schoolchildren, does the driver’s spouse have standing to sue the automaker? Even when an innovation is technically feasible, the legal infrastructure governing its use is an important precondition to its commercial adoption.
Another headwind that should not be overlooked is social attitudes and constructs. Much of post-War U.S. society – everything from single-family home architecture to city planning to dating rituals – has been shaped by the paradigm of personal car ownership. Shared autonomous vehicles, functioning within a network of smart devices, will force nearly unimaginable changes to the way we arrange our lives and cities, and may require large-scale re-planning.
Implementation Timeline
Tesla (TSLA) Autopilot and GM (GM) Super Cruise are functionally Level 3. However, due to product liability concerns, neither company advertises products using these systems as such, but rather market them as Level 2.
GM has announced that in 2018 it will begin commercial testing of Level 4 autonomous vehicles with its joint venture partner Lyft. We also foresee increased use of Level 4 autonomous vehicles in well-circumscribed public transportation settings such as amusement park or airport shuttles. Conversations with industry sources lead us to believe that several automakers will begin marketing high-end models with optional Level 4 systems by 2020.