Earlier this year I spoke with Lior Ron, ex-Googler and co-founder of startup Otto about his plans to let trucks drive themselves on US highways. Otto has a mix of high-calibre engineers from Apple and Tesla and its goal is to improve road safety and the lives of truck drivers. But now Uber has acquired Otto and has called the partnership with Otto the “dream team”. In its blog post, Otto plans to work with Uber to make “transportation as reliable as running water”. Part of Otto’s speciality is its use of Lidar technology, which is similar to radar but provides a more detailed view of surroundings.
I catch up with Richard Wallace, director of Transportation Systems Analysis, Center for Automotive Research in Michigan, US by phone to get his views on the Uber-Otto deal, the uniqueness of Lidar technology, and the state of the autonomous vehicle space in general.
What do you make of Uber's acquisition of Otto?
I think it has a number of advantages for Uber. First they get Anthony Levandowski, one of the leading autonomous robotic engineers. He worked on Google’s first self-driving car, developed an autonomous motorcycle some years ago and has made his name in this field. That's why Google acquired him many years ago. It allowed them to jump-start their automation programme. He left Google something like a year and a half ago to set up Otto which focuses on autonomous trucks. They make their own Lidar technology, so to some extent it's a three for one deal for Uber. They get an individual who is remarkable in this area, they get a company with advanced automation technology, and they get their own internal source of Lidar technology. They don't have to go buy it from somebody else. You saw earlier this week that Ford and Baidu are making a big investment in Velodyne which is another Lidar company that has been around for quite a while. Uber does not have to go and find somebody like that to source it - they will have it internally which is a great advantage for them.
At Uber they have been really focused on ride-hailing services and all the talk about the automation has been around getting the driver out of the picture as that's a major cost for them to do business. If they didn't have to pay somebody to pick you up that would be a real motivator for them because it can increase their profits. But this opens up other transportation avenues from freight movement to other fleet opportunities that Uber was not well positioned to do before.
What about Otto - what will they get out of it?
Well usually they get money right [laughs]. They get a larger home and go from being a startup to being part of a pretty established company with Uber. They get additional resources that entails additional test facilities and labs, like Uber’s major facility in Pittsburgh for example. I think together they will both benefit.
What is so special about Lidar technology? Why are the auto-manufacturers keen on it?
Elon Musk has famously said that Tesla can have self-driving vehicles without using Lidar. Technically speaking he is probably right. You probably could do without. You and I drive cars and we don't use Lidar – our cars don't come with that technology. Lidar has some really good advantages over just using cameras. It is a constant scanning technology. It's very good with ranging like radar – it can tell how far away another object is, like another car or even a pedestrian. Radar doesn't tell you the size or shape of anything – it just tells you there is something out there. Every object becomes like a dot. Whereas Lidar gives you an overall sense of the shape and size much like a lot of vision systems can do. It's a pretty powerful tool. The Tesla vehicle that had the fatal crash in Florida – if it had been equipped with Lidar, we don't know for sure – there is good chance it would have detected the truck that is crashed into.
But what's discouraged people from using Lidar is that it is extremely expensive technology and the Velodyne tech in particular has been quite expensive. Other companies have started to bring the prices down and there are startups out there that are working on other Lidar configurations. I know Continental brought a Lidar company a few months ago and is incorporating that into their products. I think for Uber - they don't have to chase suppliers anymore. They just brought it inside their hold by buying Otto.
Uber has struggled with regulatory challenges in numerous countries. How will they move forward in this area?
Well those hurdles have been fairly specific to their ride-hailing services where they mirror or compete with taxis. In these segments like delivery and heavy vehicles they may not have the same regulatory issues. They may have others perhaps, the same ones that anybody developing automotive technology will have to overcome like licensing rules in the states or countries. That doesn't make them any different from Google or Ford for that matter or Toyota. But in terms of how they operate as a ride-hailing service and how that interacts with the taxi type laws – this opens up markets for them where they don't even have to play that game because they will be in the freight movement business – it will open up other regulations of course but it gets them out of that taxi competition. I think they will still continue to go after those `markets too – this will just broaden their scope and offer them different lines of business to operate in.
Ford announced this week they are planning to go fully autonomous by 2021. Google is also going for an all or nothing approach. Which approach do you think is good? The semi-autonomous or fully driverless?
I think they both have their merits depending on time and place. When we have it all worked out and once the regulatory and technology pieces are in place - going all out is clearly a good solution. I don't know if the 'no steering wheel' or no back-up plan is really a good idea for a long time to come.
I think there is a lot of things to overcome on that front like regulation. The semi-autonomous space has huge territory – all the way from simple systems like electronic stability control which is almost invisible to the driver, systems that are just part of normal operations – and then all the way up to the super-cruise and auto-pilot that Tesla and others are developing – that can stretch a long way to get to a point where you don't drive at all. I think the driver's assistance realm can provide a lot of value in that area for companies that are pursuing that [semi-autonomous] strategy.
I think for somebody like Google it makes sense to go for the all or nothing approach because they don't have any legacy market or legacy vehicles – they can plan ahead for that when they are ready to go all or nothing. For Ford I suspect it requires a few parallel pathways because they are a car company – so they are going to have makes and models to compete in the market with GM, Toyota and Honda and others that aren’t all or nothing. But that's a different pathway because the big issue to go with all or nothing is not what kind of sensors or drivers assistance you have – it's what kind of artificial intelligence you have. It's almost a different wing of their research centre.
I suppose this is where the advantage of the tech companies like Google comes in? Google has the tech....
I think that's kind of right. I've long thought they have a bit of advantage when it comes to the end-game strategy because they don't need to maintain or upkeep or do warranties with existing vehicles already out on the road. They don't have to worry about next year’s model. They just keep focusing on when they are ready – then they launch and its full automation. They still have to meet the existing federal motor vehicle safety standards in various countries and currently those all tend to say there needs to be steering wheels, both pedals and crash survivability. Google has managed to avoid all of that so far because they have stuck with the below 25mph vehicles that get special exemption under US law.
Let’s assume Ford is realistic in its estimates for 2021 – they might choose to put those in the market place under very controlled conditions. Like they will only offer them to certain fleets under lease deals that allows Ford to recall them if they have to or restrict how they can be resold. Ford may choose to only sell them to Lyft or FedEx or something like that.
Many car-manufacturers are teaming up. Any guesses on who is next?
There have been so many that I won't be surprised at all if this continues. I don't think there is an end to this. I think these technology companies and traditional auto companies are going to continue to find synergies where they operate together – it seems inevitable. You know, it's also not that new. Either the OEM would go and buy companies with good technology or the tier-one suppliers would go and buy them to bring them into the market place. I think maybe the amount we are seeing is a little bit unprecedented but it’s been going on for a hundred years! [laughs] There's a few other Lidar startups out there – is someone going to try and buy them? Very interesting sets of acquisitions out there.
NEXT ARTICLEGRC veteran says risk and reward must be balanced»
Adrian Schofield sheds light on tech in South Africa
Mark Chillingworth on IT leadership