Lost in the swirl of negative publicity around Uber offering rides during last week’s taxi strike in New York City (held in protest of the Trump Administration’s executive order temporarily banning travelers from seven predominantly Muslim countries), was Uber’s intriguing deal with Daimler to add Mercedes vehicles to its fleet. While Uber CEO Travis Kalanick was opting out of a Friday meeting of President Trump’s CEO advisory board, his company was touting its third deal with a car company around developing self-driving cars.
Uber would like you and its deep-pocketed investors to believe that it has the jump on the automotive industry and Waymo and all of its other competitors in the race to achieve autonomous vehicle operation. The latest deal with Daimler, for Mercedes vehicles to be integrated into the Uber fleet, positions Uber’s network as a real-world proving ground for self-driving cars.
Why test cars on tracks if you can test them in real world circumstances? This is doubly important given the estimates of statistical experts suggesting that perfecting automated driving will require the collection of driving data from hundreds of millions of human-guided, machine-driven miles. (Tesla Motors now claims to have 1.3B miles of such collected data.)
The Uber story becomes even more compelling when one considers that Uber is already working on self-driving car technology with vehicles from Ford Motor Company and Volvo Cars. There’s just one little problem in that the Ford-based testing in California is reputedly only for map data gathering, according to Uber; while the Volvo tests are limited to Pittsburgh and Arizona.
The Uber proposition might be a game changer, if Uber could find a way to enable Tesla Motors-like automated driving in combination with Uber’s ride hailing service. Instead, Uber has grafted ride hailing onto Google/Waymo-like self-driving car technology (reportedly in Phoenix). This fails to solve the highway driving challenge, as self-driving Uber vehicles, like Waymo’s, will be limited to lower-speed secondary and urban use cases – which is okay. Uber is using drivers in self-driving-enabled Volvo XC90's in Pittsburgh for a wider range of driving scenarios.
Tesla Motors’ autopilot-equipped vehicles, in contrast, are best suited to highway use, with Level 2 automated driving requiring that the driver remain engaged. The secondary and urban settings favored by Waymo and Uber remain a blind spot, for now, for Tesla, which is nevertheless gathering way more useful data in greater volume than either Waymo or Uber.
The challenge is to marry the commercial application of ride hailing as well as car sharing with the automated driving technology. Tesla has indicated its own car sharing/ride hailing plans on the drawing board – under the control of Tesla. It’s a tantalizing prospect as more car companies enter the car sharing business – building networks of cars that can be used to gather the driving data necessary to build automated driving systems.
The fact that Daimler turned to Uber clearly reveals that many car companies lack the resources and expertise necessary to manage the logistics behind car sharing and ride hailing systems including payment, routing and logistics. Car companies have also been slow to grasp the nature of the data gathering task for anything more than emergency response, roadside assistance, and diagnostic data.
Tesla is a notable exception as it sells vehicles directly to consumers and has compiled a substantial track record in managing its customer relationships directly. Most car companies deal with consumer indirectly through franchised dealers.
Driving cars with sophisticated data gathering equipment as Uber, Waymo, Ford, Volvo, Daimler and GM among others are currently doing is expensive and time consuming. But shared cars outfitted with appropriate self-driving technologies and sensors can help pay for themselves if the data can be put to work in the service of self-driving car development.
One company offers this particularly attractive value proposition. BMW’s ReachNow partner, RideCell, offers just this prospect: a commercial platform suited to optimize and maximize vehicle use across multiple applications (ride hailing, carpooling, car sharing, campus, concierge) and also capable of facilitating automated driving development.
To be clear, RideCell is merely a service provider – and a white label (ie. nameless) one at that. The company is not in the hardware business in any way and isn’t building a consumer brand. But RideCell systems are suited to gathering data in support of program logistics and may ultimately prove useful across a broader spectrum of telematics services including vehicle development and delivery and asset management.
RideCell is redefining the nature of the telematics service provider – previously defined by General Motors’ OnStar division, now known as Global Connected Consumer. In fact, RideCell’s very existence highlights the fatal flaw of Uber building its business on the backs of its independent freelance drivers.
Over time, the quality of Uber vehicles and possibly even the quality of Uber drivers can be expected to steadily decline. But that doesn’t even matter because the worst aspect of the Uber business model is all those human-driven miles with precious little data gathering. Uber’s greatest strength – not owning its assets or directly employing its labor force – is its fatal flaw.
For its part Lyft might find a way to sidestep this flaw if it can offer up automation-enabled Chevrolet Bolt’s – currently in testing in California care of GM’s Cruise acquisition. Of course the idea of a self-driving shared Maven vehicle from GM is a possibility as well. Those are a couple of big maybes.
In the meantime, car makers are flocking to RideCell to get themselves into the ride hailing/car sharing game. The concept of using Uber’s network as a proving ground for self-driving cars is attractive, but the RideCell proposition is more flexible and compelling.