Aurora recently announced two programs – Horizon and Connect – to allow partners to rent driver systems for trucks and automobiles, respectively.
The announcement is pretty high-level and thin on details. There’s not a lot of context on how, exactly, the driver gets to the customer. Does the driver come with an Aurora vehicle, or can an Aurora driver be installed in a customer-owned vehicle?
But it’s a small step toward the dream of autonomy-as-a-service. The goal is something like Amazon Web Services, whereby any person or business with a credit card can rent a small vehicle, and then a small collection of vehicles, and then progressively larger vehicles that scale up and down, on-demand.
Luminar’s $5 billion market capitalization is by far the largest of the half-dozen lidar start-ups competing for business among both automotive manufacturers and autonomous vehicle technology companies. The growth to support that valuation may well come as much from software and artificial intelligence as from the hardware sensors on which Luminar has so far built its business.
I spent a good chunk of today watching yesterday’s GM Investor Day presentation, which is available online.
I watched Mary Barra (GM CEO), Travis Katz (BrightDrop CEO), Dan Ammann (Cruise CEO), and Doug Parks (GM EVP for Product Development), all of whom were great.
Two of Dan Amman’s slides really struck me. One is the slide above, which breaks out some very high level unit economics for Cruise. Unit economics are sometimes hard to come by in the autonomous vehicle industry, and I love that Dan put them out there.
I also like this slide that predicts Cruise’s ramp, in terms of new vehicle additions.
Presentations are great, but numbers are even better. Kudos to Dan Ammann for bringing the numbers.
Walmart announced today a partnership to deliver online purchases for Home Depot. This is the first partnership in Walmart’s new GoLocal white-label delivery service.
On the one hand, this is another salvo in the Amazon vs. Everybody retail competition. All of Amazon’s competitors, which include both Walmart and Home Depot, have to hang together.
On the other hand, this seems like a really savvy move by Walmart to become the delivery network for retail. Uber and Lyft already do this for people, DoorDash and Uber Eats already do this for food, but nobody has yet done this for packaged goods. Arguably Amazon does this in a sense, with its “Everything Store” and third-party marketplace.
But Walmart still has a pretty wide open field to aggregate suppliers. The real competitive advantage seems like it would accrue if Walmart were to aggregate customers. If GoLocal were to become a destination website, that would further motivate suppliers to join the platform “on Walmart’s terms” (to paraphrase Stratechery’s Ben Thompson).
That has some serious implications for autonomous vehicles. Argo has already chosen to launch on Lyft, because that is where the ridesharing customers are. If Walmart succeeds with GoLocal, perhaps that is where the delivery customers will be.
Qualcomm will pay $4.5 billion to purchase Veoneer, a Swedish Tier 1 automotive supplier with a specialty in advanced driver assistance software. That makes Veoneer one of the largest acquisitions in the autonomous vehicle space since Mobileye. Interestingly, both of those companies focus on ADAS, although Mobileye seems to be stepping into full autonomous driving, as well.
There was a little bit of a bidding war in which Qualcomm beat out Magna on the purchase price. The deal got relatively little attention in the press, however, I think because Silicon Valley still has not really figured out how to break into the automotive supply chain. Qualcomm, Mobileye, and Magna are all located outside of Silicon Valley.
But there’s big money in ADAS, as evidenced by first Mobileye and now Veoneer, to say nothing of Tesla.
Qualcomm’s head of automotive, Nakul Duggal, appeared recently on The Autonocast, in a short episode that’s worth a lesson. Among other things, Qualcomm does not aspire to be a Tier 1 automotive supplier.
A few weeks ago, Apollo announced a joint effort with Baidu’s DeepWay subsidiary to launch a semi-autonomous truck called Xingtu. This marks a further expansion of Apollo-powered vehicles, from tiny delivery robots to public safety patrol vehicles to robotaxis to public transit shuttles to buses, and now to trucks.
Autonomous trucking shares many of the challenges of autonomous urban driving, but trucking avoids some problems that robotaxis encounter, while experiencing different hurdles.
For example, autonomous trucks can mostly focus on operating in controlled highway environment, without the cross-traffic and chaos that robotaxis have to handle in urban environments.
On the other hand, trucks operate at much higher speeds, and thus have to see much further down the road, in order to react and even stop before a collision. Apollo states that the Xingtu vehicle can perceive objects up to one kilometer ahead. Long-distance perception typically requires a trade-off between distance – whether that comes from a special camera lens or high-powered lassers – and field of view.
Trucks also are much heavier, and have much different vehicle dynamics. Robotaxis are typically so-called “rigid body” vehicles, whereas truck as “articulated” – the rear trailer can move somewhat independently of the front cab.
Nonetheless, enough of the core autonomous vehicle technology overlaps that it makes sense for Apollo to pursue trucking. The trucking market is even more important outside of China than within it, due to the highly urbanized nature of China’s population distribution. In that sense, the expansion to trucking could be a harbinger of Apollo’s growth beyond the Chinese market.
Apollo’s move to trucking parallels Waymo’s deployment of their Driver technology to both autonomous trucks and robotaxis. In a recent CNBC interview, Waymo co-CEO Tekedra Mawakana highlighted the ability to deploy the Waymo Driver to both their Waymo Via freight service, and their Waymo One transportation service. Pushed to pick one area over the other as a focus, Tekedra countered that both services depended on a common platform.
Cruise Infrastructure VP Vinoj Kumar appeared recently on the Software Engineering Daily podcast to talk about the cloud computing that support Cruise’s autonomous vehicles and forthcoming customer-facing products.
“For example, AV’s need to securely connect to backend services, the broker communication to a set of microservices that handle everything from dispatch, remote assistance, mapping, role planning, etc., right? And once the drive is done, there is more data movement. For example, ingesting data into our data lakes for post processing, then you do analysis, continuous improvement.”
Artificial intelligence requires tremendous amounts of data, which necessitates world-class data infrastructure. Self-recommending!
Motional, the autonomous vehicle company formerly named nuTonomy, has traditionally been the most quiet of the major US-based AV startups. Although its size, in terms of headcount, is on par with Waymo, Cruise, Argo, and Aurora, Motional winds up in the news a lot less.
Recent Motional announcements have been pretty low-key, but they are at least a reminder that the company is progressing.
“Motional, the autonomous vehicle company that is a joint venture between Hyundai and Aptiv, is growing its presence in Las Vegas as it gears up to launch a commercial robotaxi service in 2023.”
The firm is making a huge investment in a new testing center in Las Vegas, as well as rolling out the Hyundai Ioniq 5. That vehicle, an SUV, will be the centerpiece of their initial ridesharing effort.
The Motional partnership with Lyft means that Lyft will have at least two self-driving partners – Argo and Motion – plus whatever happens with the former Lyft Level 5, now part of Toyota’s Woven Planet subsidiary.