Richard Bishop on the path to autonomous driving

Richard Bishop on the path to autonomous driving

As a precursor to our forthcoming webinar on Automated Driving and its accompanying market trends analysis feature, Intertraffic sat down with Richard Bishop, the man widely recognized as one of the world’s most knowledgeable autonomous vehicle experts, to explore how we got here… and where we’re going.

Richard Bishop has never been far away from any of the most significant moments in the development of driverless vehicles over the last 30 years. Formerly program manager at US Department of Transportation, in 1998 he launched his own consulting firm that has seen him work with some of the world’s market leaders and innovators such as Toyota, Peloton and Plus and seen his expertise become in ever-increasing demand, taking him around the world to congresses, conferences, seminars and workshops, culminating in his recent addition to the Forbes expert roster, but we started by asking Richard to take us back to the beginning of his time in the AV world.

Intertraffic: Richard, where and how did you first get into this fascinating sphere?

Richard Bishop (RB): "It was the best part of 30 years ago! I had had a career in the US Federal Government in the Defense Department that I had just kind of grown tired of. I wanted to do something else a little more civilian-oriented so I looked around for something within the Government and here was a job with the Federal Highway Administration (FHWA), and what was then called Intelligent Vehicle Highway Systems (IVHS). I thought, ‘Oh, this looks cool’. It sounded really weird for an electronic engineer to go to the FHWA, but I took a chance. I was the first “Double E” they had hired in the last 20 years, so I was surrounded by civil engineers. This would have been 1991. So what became ITS was nascent in the US and of course, the neatest things were happening in Europe. Congress has passed the bill called ISTEA (Intermodal Surface Transportation Efficiency Act) that established the IVHS program. And that was right after the Berlin Wall had come down, of course, and so there's a ‘swords to ploughshares’ sort of sentiment in Congress at that time. Someone on the appropriate committee in Congress put in a two-line clause in that piece of legislation saying, the Secretary of Transportation shall develop and demonstrate an automated highway system program. And so that because I was the only electronic engineer amongst a bunch of civil engineers, I was seen as the guy best suited to do that. And that became my program."

INTERTRAFFIC: That would be a great way to describe the word “serendipity”.

RB: "So huge, amazing serendipity. I'm really forever grateful for that and humbled. And it became a big deal at the time, we formed a consortium led by General Motors with Carnegie Mellon, and UC Berkeley and Caltrans, and Lockheed Martin. So it was a big deal. We organised the famous at the time Demo 97 in San Diego, which you can still find on YouTube. And, you know, I think we showed the world that automated driving wasn't just something in science fiction books, it really could be done. So we sort of shifted the mindset - we thought over the next few years we’d go ahead and develop a robust system description so that industry could take it from there. But at around that time political pressure in Washington regarding budgets meant that our long-term oriented, futuristic research program couldn't withstand the scrutiny and so they shut it down. At that point, I was restless again. So that's when I jumped off and started consulting and really enjoyed sort of putting my arms around the whole industry and trying to be useful."

INTERTRAFFIC: Going back to the mid-late 90s briefly what was the technology that was being used to automate driving at that point?

RB: "It was camera and radar. LIDAR was not in the discussion at all at that point. We were focused, we had several different technical areas of focus. One was the idea of car platooning. So that traffic could flow better, you get more capacity on the highway. So vehicle-to-vehicle communication was absolutely key to that. So it was basically a radar and camera. We had Carnegie Mellon working on a pure perception approach. So that was camera radar. We had UC Berkeley working on a more infrastructure-supported approach, so that was the idea of magnets in the roadway you experienced in Leiden and they were for tasks such as lane keeping, particularly important with the idea of snow covered roads, and that kind of thing."

INTERTRAFFIC: In 2007 you led a team in the DARPA Urban challenge. So how did that come about? It wasn’t all that long since your DoT program was cancelled and here you are heading up a team that took part in a globally famous autonomous vehicle competition.

RB: "It was around 2000 when the connected vehicle concept caught fire. And everybody's really excited in the US, we had the spectrum allocated in 1997, so that was the real focus of the ITS community. Back in the 90s, I had been in touch with my DoD counterparts who were doing unmanned ground vehicles for scouting and supporting the troops and that kind of thing, so they kept their program going and came up with this innovation of a challenge. So that instead of writing a statement of work and putting out an RFP, they just said, ‘Hey, here's a million dollar prize if you can do this thing’. And that really motivated and stimulated a whole new cast of characters, universities, etc. So the first challenge was a desert challenge. The vehicles weren't too successful. A couple of years later the vehicles did much better and that subsequently led to the urban challenge. This was, as you say, 2007, and it just lit a fire under the technical community and it was also when LIDAR really, really came along although it had been more widely used in some of the European programs like PREVENT. The team I lead was called Team Lux and we were sponsored by Ibeo, the LIDAR company, and I was pleased to lead his very small team. We did pretty well – I believe we were in the top 20 but they didn't rank all the entrants, but it was very impressive for such a small team and it allowed the team to show what LIDAR was capable of."

INTERTRAFFIC: So do you think that was the catalyst for this whole recent rash of technological development?

RB: "Actually, I do. I think the DARPA Challenge was a trigger point because it invigorated all these folks. But during that same decade we had the dotcom boom and crash around whatever it was, 1999. And, and so everyone was excited about startups. But I actually worked with a guy to try to do a startup to retrofit cars with adaptive cruise control. And we got nowhere because that startup world was all about software. If there's any hardware involved, forget about it. But the startup culture existed, and the DARPA Challenged kind of came together with that software-based startup culture, all because of Google. Google, the founders, Larry Page, and Sergey Brin, said ‘Hey, this is cool’. The startup mentality is one in which it's cool. And you're doing pretty well, you can just give it a shot. There's a story that I have no idea if it's true, but I like it. So I'm going to tell it, and I'm going to run with it, which is that Larry Page of Google heard about Demo and was fascinated at the time, thought that was the most amazing thing ever. And it's that seed that stayed in his head. Page had a big company with lots of money. He grabbed some of those guys from the winning teams in the Challenge and started an automated driving activity."

INTERTRAFFIC: I suppose once Google gets involved, the game changes?

RB: "What’s cool about that is that everything up to that point had been funded either by the car companies, with their research budget, or with public money. It was mainly motivated by public money, because it was kind of visionary. So the technology was happening, moving at a decent pace for research, but not a not a fast pace. So the paradigm completely shifted, when you had a startup culture saying, let's just go do this. Instead of it was motivated by the idea of ‘well, let's prove out one function, one use case, which is automated driving’. Well, all car companies have to do is sell cars to continue to exist, so there's no forcing function. When Google got into the game, it wasn't what I call a forcing function as such, but it was this opportunistic mindset that just changed everything. And then when Google first came to the public and revealed that in 2010, they'd been running automated cars on public roads, it could have been a huge negative public backlash. Who knew how that was going to go? But the public was intrigued and excited. Car OEMs would have risked a lot to do that, whereas Google tested the water for them. And so the OEMs shifted gears during the 2010 to 2020 decade, and what we're seeing is what has emerged from that combination of factors."

INTERTRAFFIC: In 2014 Chris Urmson, who was then heading up Google’s self-driving car project, said that balance of human-driven vehicles would drop below autonomous vehicles around the time his then-12 year old son was old enough to drive. Was that an overly ambitious statement or was it realistic and in which case did something unforeseen occur that hampered technological development? Whatever it was, that obstacle has been well and truly surmounted or circumvented.

RB: "There was that kind of talk back then… and through 2015 to 2018. Every decade is like a marking point - everybody was saying 2020, 2030 but I heard that his introduction dates for Google were 2018 for their first robotaxis. We’re kind of there for Waymo, which is what spun out of Google, they have launched fully driverless services to the public. The difference here is that the people who follow automotive journalism were used to how OEMs talked about things. And then they introduced a product that had a radar on it, that meant it was available. It's very specific geographically, and very limited in function, but it's still got a useful function and a useful use case. I've been asked the question of when, over and over and over again. The actual question should be ‘when and how much?’ as that's when it starts to matter. We've got the first driverless robotaxis now. They will scale modestly over this coming decade. But on the truck side, where I'm very active, the business case is so strong there. We’re looking at 30-40% reduction in cost to move freight! That could scale a lot quicker. They don't have the first driverless vehicles out there at this time, but they're pretty close within a couple of years."

INTERTRAFFIC: Long-distance trucking always seemed like the most viable route to mass market acceptance for autonomous vehicles, especially looking at let's say, Australia, where truckers have to transport goods from Perth to Sydney on one road that stretches for several thousand miles the country. You are hugely involved in the trucking side of things, so was that a bit of serendipity that you got so entrenched into that or was that a deliberate move on your part?

RB: "Oh no, it was serendipity, just by me circulating amongst all that was going on. I was actually supporting Auburn University in Alabama and their research efforts and I still am. And the professor there, Dr. Beverly, knew the CEO of Peloton technology because they'd gone to Stanford together. We started working together, and it just grew from there. In terms of the investor community, it's funny, because the original buzz was about passenger car activities rather than trucks, but it seems like in the last three years or so, the investors kind of woke up and thought ‘Wait a minute – trucks! Wow, that's where the return on investment looks a lot better!’ And you're not trying to create a new business called robotaxis, this is an existing business where you're replacing driver with automation. Now, in the COVID era, trucking looks a lot stronger to the investors than the other use cases."

INTERTRAFFIC: So what would what would it need for that to become a reality in the next five to 10 years? Will it need a global haulage firm, a FedEx for example, to plunge headfirst into automation? Is the freight sector waiting for a company of that size to put its hand up and be the first or is this a technology that maybe isn’t entirely ready yet?

RB: "That is one pathway forward and I would say it's the most likely. There are also trucking companies like Kodiak that just develop the technology then create a fleet of autonomous vehicles and start moving freight around. So it's possible for this to start on its own, independent of an existing fleet, but UPS has been visibly active in this space, FedEx a little bit, and a few others. You know, they're watching very closely but I think you’re right, it will be triggered in a large way by those guys."

INTERTRAFFIC: Where do the major truck OEMs fit into the scenario, though? We’re talking about startups and innovators here, but what about the likes of traditional truck manufacturers such as Mack or MAN?

RB: "A really important turning point with the truck OEMs was in the last year or so. The startups could buy a truck and they hack into them and apply their software to figure out how to do this robustly. And that's fine to get started. But the time came when they had to get more and more integrated with the truck, the innards of the truck, the electric power system and the brakes and engine, so what's happening now is a much deeper integration between the startups and the old guard coming together. They know they need each other to make this happen. And we've seen some things in the news with TuSimple working with Navistar and others, Waymo has placed an order with Daimler Trucks to deliver OEM factory-built trucks. And the key part of this is there's an infrastructure inside those trucks that is needed for level four autonomy, which is redundancy in the braking and steering plus cooling systems for the electronics and redundancy and electric power, all of that. The startups can't do that by themselves - it needs to come from the factory, so we may see some retrofitted driverless trucks in the next year or so. Small numbers, but the general consensus is 2024 is when the factory-built systems will be coming off the production line. And then that'll be another turning point."

INTERTRAFFIC: So with regard to the next five years, as 2024 is only three years away, in terms of passenger vehicle automation, where do you see that going? You're in the best place of anyone on the planet to answer this question, so where are we going to be? Are we now, in late 2020, where you thought we’d be in 2015 in automation implementation, or has the sort of the landscape changed?

RB: "We have two important threads here. One is the personally owned vehicle that you buy from an automotive OEM and the other is the fleet and robotaxi guys. So in the personally owned vehicles space we know from lots of car manufacturers that you can buy an advanced level two system that does the lane keeping and longitudinal control and high convenience factor there. Tesla does more than that but they don't do full self-driving, even though that's their words, but they are pushing the rest of the industry. They're the most aggressive in terms of function, and the customers absolutely love it. So what we're seeing is an incremental approach from the car manufacturers. Level three, the idea that the driver can be hands off, feet off and eyes off while the system takes care of driving, even though you have to leave your brain on to take over upon request. That's been delayed. That was the kind of thing that was promised originally around 2020 but didn't happen because of regulatory issues. BMW and Mercedes both indicated that level three autonomy will come at the end of 2021 and we'll see others doing that. Level four is some time off, though, and I think the OEMs are not in a huge hurry. They were at one point, there was a lot of excitement for a while, but now the attitude is that they'll get there when they get there."

"On the robotaxi side it will scale at its own pace. I don't know how fast it'll go, honestly. But we’ll see cruise automation, which has been active in the city of San Francisco, very soon going to actual driverless. So we're now at the point where we’re moving from safety driver in the automated car to a completely driverless car. So you know, the joke has been how can you tell if a car is driverless? Well, it's got two engineers in the front seat."

INTERTRAFFIC: What do you think is the next big kind of eureka moment for the autonomous vehicle sector?

RB: "That's a good question. There’s a couple of things, I'll do some random access here. You know, right now, the robotaxis are running on fairly benign streets, so they can only operate in limited geographic areas, so they can only serve a certain number of trips. To use Silicon Valley as an example, that's a place where to even take a two mile trip in an Uber, you're probably going to go on a street for half a mile, and you're going to zip up a little freeway ramp, and then you’ve got to get off at the next exit, then you go to your destination. So this transition of robotaxi capability to both highway and street is going to be really important. They've got to do that. That's part of scaling up. San Francisco is a jurisdiction that talks about the need to apply some kind of a tax or fee on ride-hailing, in general, but even on robotaxis so that fee would go up or down based on the number of passengers you had in the vehicle to try to encourage more ride-sharing of a ride, even though COVID has pushed back against that. And so that regulatory side is going to be really important. There's going to be some pioneer jurisdictions that look to implement that in Europe, probably Australia and the US."

"On the truck side I think it's the mundane things of how do you really integrate these trucks with automated capability into an overall freight situation, the broader logistics, and then even a much broader sense the supply chain. The folks who don't think about trucks, they think about goods, and we've got to get these goods to move back and forth. My favorite example is if you've got a truck that needs to go from California to Pennsylvania, that's probably a six or seven day trip, because the driver rest times. The company I'm working with, Plus, are developing automated trucks that did what they call the ‘butter run’ from Pennsylvania to California in two days because the truck didn't have to stop as there's no driver rest times involved. When you can move a load cross country in two days, instead of six days, it affects some freight that moves by air, for instance, because of the delays of trucking. For refrigerated loads, you've got a refrigeration unit, keeping all that butter cold for six days, but now for two days, you've got an environmental benefit there. I think that's just the tip of the iceberg. There's so much the automation is going to ripple through the supply chain in myriad ways. I'm really looking forward to see how that plays out.