Robotaxi services: Closer than you think
January 12, 2021 | Company Blog
At the end of last year, Alphabet/Google’s Waymo announced that it will introduce paid robotaxi services in Arizona and California—without safety drivers on board. The Arizona service launched in December without fanfare.
The emergence of paid, safety driver-free service marks a major step forward in the autonomous mobility world and shows a path that others might take to reach the milestone. General Motors’ Cruise, Amazons’ Zoox, Motional, and AutoX also received their permits to operate robotaxis without a backup driver in vehicles in the U.S., but of these only Cruise and Motional announced plans to launch commercial services in 2021. In China, Baidu’s Apollo Go and Didi Chuxing have been offering free robotaxi services (with safety drivers) since mid-2020.
How did these companies get to this point? Writing in EE Times, automotive analyst Egil Juliussen identifies five stages of autonomous service development that the pair have followed:
While Waymo, Cruise and Motional reached stage 5, AutoX is currently at stage 4, trialing free robotaxi services with safety drivers in 2019 and has a permit for attendant-free services. Apollo Go and Didi Chuxing are at stage 4. Zoox, acquired by Amazon in 2020, is at stage 2, testing vehicles with safety drivers for a limited number of users.
The emergence of paid services highlights different approaches to bringing robotaxi services to market. Waymo is focused on self-driving technology and plans to integrate it into third-party vehicles (as are Aurora, Mobileye, AutoX and others). Cruise, on the other hand, is developing its own technology and a new GM vehicle for an integrated platform (as are Tesla, Zoox/Amazon, and others). Motional, the rebranded Hyundai/Aptiv joint venture, is supplying vehicles for Lyft to operate services.
Waymo started its robotaxi initiative in 2017 (even earlier as part of the Google Car project), Cruise in 2016. It has taken this long to operate enough miles to get the data needed to train their respective software platforms to the point where they can operate safely in live traffic. Juliussen estimates a 2 to 4-year timeline to reach this point—making this estimate after Waymo and Cruises’ announcements.
A major obstacle for any business entering the AV space is the vast cost of developing and testing the technology. It is no coincidence that both Waymo, Cruise, Motional, and Zoox have extremely wealthy parents—Alphabet/Google, General Motors, Hyundai, and Amazon respectively. Massive amounts of test data are needed to deliver the information that advanced algorithms need to make the right driving decisions. The time it takes to obtain that data while offering free services makes the effort extremely expensive.
Uber’s self-driving group was said to be going through $20 million cash burn per month (the company has since sold that part of its business to Aurora). Autonomous trucking startup Starsky Robotics also shuttered in 2020, unable to secure more funding. Chinese robotaxi operators are ahead of those in the U.S. and Europe in part because of government support—investments in companies like Baidu as well as in the infrastructure (data centers, 5G networks) needed for services to scale. Also, there are no consistent regulations across states in the U.S. or among European nations, and as monolithic entity China able to faster on this front too.
There is also the question of what business model or models will make robotaxi services profitable. Will companies building software-only sell or license their technologies or will they seek to become service providers—Waymo seems headed in the latter direction. Hyundai seems to be planning on selling/leasing vehicles to service providers like Lyft. Will integrated self-driving vehicle makers make a move from selling cars to selling mobility-as-a-service directly to travelers?
According to Urban Mobility Daily, cruise parent GM and the other automakers are at a “now or never” moment, citing a recent conference where Nissan’s Ivan Ollivier suggested automakers should “forge partnerships and work with entire systems – cities, regulations, etc.” to create franchise-type relationships. This would make robotaxi services akin to today’s public transportation services, with cities controlling supply and demand to avoid the over-proliferation of vehicles that happened with the arrival of Uber and Lyft.
The city partnership model may also be needed to ensure that new autonomous services mesh well with other modes of transit to reduce congestion and make Mobility-as-a-Service (MaaS) as convenient as possible. A seamless blend of modalities—from bikes to robotaxis to buses to trains—enabling door-to-door trip planning with predictable schedules and connections is what will ultimately make it easy for urbanites to give up driving. This will also require data sharing among all of the entities involved in delivering MaaS—from traffic to weather to timetables to real-time capacity so that travelers can make informed decisions when planning trips.
Robotaxis services, on-demand autonomous ride-hailing, has long been seen as the endgame of driverless technology. Shared, electric, autonomous mobility services promise to reduce private auto use—and, with it, congestion and pollution—on city streets. Fleets of cars that are smart enough to drive themselves and pick up and drop off passengers anywhere in town 24/7 can move more people with fewer vehicles at a far lower cost for residents than buying, maintaining, and storing a private car. That’s the theory, anyway.
Much remains to be seen about the viability of robotaxi services. The quality of service and its integration with other transport modes are sometimes overlooked in the quest for safe self-driving technology. Self-driving tech is just one component of a viable service, and services will also have to mesh with the overarching mobility ecosystems.
Travelers around the world, especially commuters, are addicted to convenience. This helps explain why so many pay some $10,000 a year in the U.S. (just £5744.40 in the U.K.) for a personal car that they use just 5 percent of the day. For robotaxis to be as or more convenient than driving will require intelligent fleet orchestration for every service—automated dispatching, routing, and ride matching that always sends the right vehicle to the right place at the right time with predictable efficiency. The good news for advocates for safe autonomous mobility is that progress continues to be made in self-driving technology and in developing viable services, with large-scale deployment now realistically in sight.