Views from the Marketplace are paid for by advertisers and select partners of MIT Technology Review.
Autonomous Vehicles: Are You Ready for the New Ride?
Automakers are joining with Google, Uber, and high-profile start-ups to harness the technological advances that will power next-generation autonomous vehicles.
The self-driving car revolution is about to shift into overdrive. The signs are everywhere. Just look in the back lots of South Boston’s tech corridor, on the streets of Pittsburgh, in the prefab test facilities at the University of Michigan’s Mcity, and throughout a smattering of open highways.
Car companies are joining with tech giants like Google, Uber, and prominent start-ups to develop next-generation autonomous vehicles that will alter our roads and throughways and lay the groundwork for future smart cities. They’re harnessing technological advances such as machine learning, Internet of Things (IoT), and the cloud to accelerate development.
More significantly, autonomous vehicles will advance the industry disruption set in motion by popular ride-sharing services like Uber and Lyft. The pieces are coming together to create a world where intelligent, driverless vehicles become the future of transportation.
“Autonomous vehicles will help bring the city back to what it was—for people,” says Ryan Chin, co-founder of Optimus Ride, a Boston-based startup working on self-driving technologies. As autonomous vehicles gain traction, Chin envisions an opportunity to remake the city landscape by consolidating parking, reclaiming land for parks, reducing urban congestion and traffic, and promoting overall highway safety. “Autonomous vehicles will play a big role in delivering a much safer environment because they follow the rules of the road,” he says.
While much of the early hoopla has focused on the still-formative vision of fully autonomous vehicles (those that attain Level 5 as defined by the Society of Automotive Engineers), there are different levels of self-driving capabilities. Some Level 1 features such as adaptive cruise control, automatic emergency braking, automated parking, and active lane control are already mainstream features in current vehicle models. Luxury brands including Volvo, BMW, and Mercedes-Benz have begun to showcase Level 2 features such as automated steering and speed control for short periods of time. Level 3 cars (still in testing) will require some driver intervention. Level 4 will be fully autonomous but can still be driven by humans, and Level 5 vehicles will be designed to take the driver completely out of the picture.
Eventually, all self-driving cars will employ some combination of sensors, cameras, radar, high-performance GPS, Light Detection and Ranging (LIDAR), artificial intelligence (AI), and machine learning to achieve their respective levels of autonomy. Connectivity to secure and scalable IoT, data management, and cloud solutions are also important to the mix, providing a resilient and high-performance foundation on which to collect, manage, and analyze voluminous sensor data.
While the era of the connected car is still in its infancy, there is widespread optimism that it’s the wave of the future. Gartner projects that 250 million connected vehicles will hit the road worldwide by 2020. Moreover, the World Economic Forum anticipates driverless vehicles will generate $1 trillion in “economic benefit to consumers and society” over the next 10 years, and autonomous driving features will help prevent 9 percent of accidents by 2025 with the potential to save 900,000 lives in the next 10 years. By 2040, autonomous vehicles are expected to comprise around 25 percent of the global market.
The rise of the connected vehicle has far-reaching societal implications, from environmental benefits to improved safety. Fewer cars on the road means a reduction in greenhouse gas emissions, leading to better air quality and lower energy consumption. Beyond sustainability advantages, self-driving cars are poised to open up a whole new economic chapter in what Intel and research firm Strategy Analytics are calling the Passenger Economy. In their economic vision of the trend, valued at $7 trillion in revenue by 2050, drivers will become idle passengers and thus potential consumers of new goods and services such as onboard beauty salons, health-care treatment pods, and multimedia content primed for on-the-go consumption.
Autonomous operation will also change the public’s relationship with cars, as vehicle ownership takes a back seat to Mobility as a Service (MaaS), where individuals make use of self-driving cars on demand much like any other utility service. The Intel/Strategy Analytics report estimates that business use of MaaS could generate $3 trillion in revenue by 2040 and consumer use could account for $3.7 trillion in the same time frame.
Software will be the key enabler of autonomous vehicles, allowing for instantaneous new-feature updates, but also serving as the mechanism for personalizing the experience and programming the vehicle’s performance. For example, families could configure a self-driving car to do continuous pick up and drop off, shuttling mom off to work and carting kids to school and soccer. With this scenario, MIT estimates autonomous vehicles could meet society’s personal transport needs with 80 percent fewer vehicles in transit.
There are also significant ramifications for safety and traffic concerns, with experts predicting far fewer accidents. Self-driving cars are projected to save 585,000 lives between 2035 to 2045, according to the Passenger Economy research. They are also expected to slash public safety costs related to traffic accidents by more than $234 billion in the same period. At the same time, autonomous vehicles should free up more than 250 million hours of commuting time annually for consumers, especially those living in congested cities.
As more autonomous vehicles hit the streets, city planners will accelerate plans to modernize highways and thoroughfares with smart technology for road signs, traffic lights, and merge lanes—all in an effort to reduce congestion and increase public safety. “For decades, we thought we’d get to smart highways by making highways smart, but there’s been a huge chicken-and-egg problem,” says David Tennenhouse, chief research officer for VMware. “With autonomous vehicles that can navigate existing infrastructure and talk to each other as they do [through vehicle-to-vehicle communications], we can get more efficiency out of roads. Then we will ultimately have smart infrastructure to go with the smart vehicles, but this gets everything started.”
For both the self-driving cars and the smart roadway systems, endpoint telemetry, smart software, and cloud are essential enablers. The onboard cameras and sensors on an autonomous vehicle collect vast amounts of data, which must be processed in real time to keep the vehicle in the right lane and operating safely as it heads to its destination. There’s a lot of local data processing that has to occur in real time, including the computations necessary to keep the car in its lane. At the same time, there are processing tasks that can happen remotely in the cloud, such as software updates and upgrading learning models. A scalable, highly-resilient cloud-based infrastructure is critical for handling this type of large-scale data processing, while cloud-based data management systems and intelligent agents take charge of aggregating and analyzing the real-time telemetric data—for example, vehicle speed and surrounding car proximity--to initiate actions like braking or switching lanes.
Cloud-based networking and connectivity is another important part of the mix. Autonomous vehicles will be outfitted with onboard systems that support machine-to-machine communications, allowing them to learn from other vehicles on the road to make adjustments that account for weather changes and shifting road conditions such as detours and in-path debris. Advanced algorithms, AI, and deep learning systems are central to ensuring that self-driving cars can quickly and automatically adapt to changing scenarios.
Beyond the specific components, scalability of cloud computing infrastructure along with intelligent data management and transmission capabilities are indispensable for ensuring all of the right information is processed properly and securely. This is especially true for destination and address data, which could be considered personally identifiable information. For example, built-in intelligence could determine if data storage and analysis happens in the cloud or onboard the vehicle in the event that the travel path crosses regions with sub-par network connectivity.
“These are data center problems,” says Tennenhouse. “The scale of the problem plays into the need for infrastructure and data management and for securing and managing the flow of data. That’s where VMware can truly add value.”
Cruising with Optimus Ride
Back in Boston, Optimus Ride is hard at work on full Level 4 autonomy, not for a specific vehicle it is developing but as a mobility-on-demand system that supports electric vehicle fleets. Beyond safety and sustainability, the mantra behind Optimus Ride’s still stealth-mode autonomous vehicle development is what it dubs “equitable mobility,” according to Chin. “There is a deep correlation between wealth and mobility access, and we want to change that by providing autonomy and shared transportation,” says Chin. “If you can deliver mobility access, you provide both economic and societal benefits for everyone. We want to democratize mobility for the masses.”
The MIT Media Lab spin-off is working on a full-stack solution that encompasses hardware, software, and deep machine learning capabilities, among other technologies. The idea is to offer up its stack as a white-label solution to companies interested in delivering on-demand mobility services (think taxis or ride-sharing offerings) to meet society’s changing transportation needs, according to Chin. Target customers could be cities, public transit systems, rental car companies, ride-sharing services, or even big corporations that want to provide mobility capabilities for employees, he explains.
In addition to regulatory challenges, companies in the autonomous vehicle space are grappling with a range of technical problems, including the need to create the infrastructure to handle the vast amounts of data and process it at scale. “Data velocity is a key challenge everyone has,” says Chin. “When you have a fleet of thousands of vehicles, getting good data and being able to process it through simulation is really key.”
While this spate of advanced technologies is providing an on-ramp to autonomous vehicle development, Chin, Tennenhouse, and other experts admit it will be a while before the self-driving revolution forever alters the transportation landscape. “The final frontier for autonomy is urban driving under extreme weather conditions, and that’s going to take some time,” says Chin.
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today