The future of autonomous vehicles

An authority on autonomous robotic systems talks about the impact of AI on his field here on Earth and growing aspirations for autonomy in space.

author avatar

29 Oct, 2024. 18 min read

AI is revolutionizing our ability to train cars to drive safely and to respond to unexpected events. | Shutterstock/Scharfsinn

AI is revolutionizing our ability to train cars to drive safely and to respond to unexpected events. | Shutterstock/Scharfsinn

This article was first published on

engineering.stanford.edu

Returning guest Marco Pavone is an expert in autonomous robotic systems, such as self-driving cars and autonomous space robots.

He says that there have been major advances since his last appearance on the show seven years ago, mostly driven by leaps in artificial intelligence. He tells host Russ Altman all about the challenges and progress of autonomy on Earth and in space in this episode of Stanford Engineering’s The Future of Everything podcast.

Transcript

[00:00:00] Marco Pavone: In terms of applications, now we see that autonomous systems are now becoming a reality. So we have robo taxis providing service in some cities in the United States. Unmanned aerial vehicles are now used ubiquitously and even in space exploration. Autonomy, particularly in the context of, uh, orbital applications is making a key strides. So very exciting time.

[00:00:31] Russ Altman: This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. If you're enjoying the show, please consider rating and reviewing it. We like to get fives when we deserve them, and we would really appreciate it. Your input will help spread the news about the show, and will help us improve it.

[00:00:47] Today, Marco Pavone from Stanford University will tell us how autonomous vehicles, self-driving cars, and self-driving space vehicles are improving. AI has helped a lot, and there's been great technical advances. It's the future of autonomous vehicles. 

[00:01:04] Before we get started, another reminder, to rate and review the podcast, if you're enjoying it. That'll help all of us.

[00:01:18] You know, seven years ago, I had a guest, Marco Pavone, come on to The Future of Everything and tell us what the status was in 2017 about self-driving cars. Well, I have Marco back today and he's going to give us an update on where we are with self-driving cars, and also he'll tell us about self-driving space vehicles, which he's also working on.

[00:01:38] It turns out that AI has not only revolutionized our ability to write memos and emails. It's also revolutionized our ability to train cars to drive more safely and, while they're driving, to respond to unexpected events. Marco will tell us about that and he'll also tell us that some of these technologies are now very relevant in space. There's been a blossoming of new companies with new applications of space vehicles, some of them as small as a shoebox, to monitor the Earth and help with logistics, with clocks, with positioning, of course, and with many other applications that you might not expect. 

[00:02:16] Marco Pavone continues to be a professor of aeronautics and astronautics and of electrical engineering and computer science at Stanford University. He's running an exciting new center for space and car autonomy. And he's an expert on the technical challenges of getting these vehicles to work safely and effectively for all of us. 

[00:02:37] Marco, you were on The Future of Everything seven years ago, uh, time flies. What are the big changes in the area of autonomous vehicles, self-driving cars, and I know you're also looking into aeronautics.

[00:02:50] Marco Pavone: Well, first of all, thanks for having me. Well, a lot of things have changed, both in terms of the tools that we use to design autonomous systems and in how autonomous systems are now becoming a reality. In terms of tools, of course, uh, everybody knows about the incredible advancements in the field of AI. And obviously, autonomy engineers have been very eager to use those tools in order to design in a completely novel way, how we, um, architect autonomous systems. And in terms of applications, now we see that autonomous systems are now becoming a reality. So we have robo taxis providing service in some cities in the United States. Unmanned aerial vehicles are now used ubiquitously. And even in space exploration, autonomy, particularly in the context of, uh, orbital applications is making key strides. So very exciting times. 

[00:03:51] Russ Altman: This is great. This is great. And so let, let's go, you, you've just hit so many good topics. Let's go to the, um, the self-driving cars, because seven years ago, you know, there was a huge bubble of enthusiasm at that time.

[00:04:04] And I'm not going to say any bubble was popped, but, um, uh, then reality kind of set in and people were very worried about safety and trust. And you and I talked about that back then. What is the status of the, um, enthusiasm for fully automated vehicles versus some kind of hybrid? I think we discussed back then there's different levels of automation, of automatic, um, autonomous vehicles.

[00:04:30] Um, where are we, um, the robo taxis do seem to be absolutely autonomous. Um, there was talk about trucks coming down the interstate. And then there was the idea of every individual having a car that could drive around. Have we changed our expectations or are they, are we still shooting for the same goals?

[00:04:49] Marco Pavone: Expectations have changed for sure. Uh, as you said, um, as every technology, the, um, self-driving technologies had its up and downs. Uh, in the 2018, 2019, I will say was probably on the low side, as you said, realities, uh, kicked in and, uh, this technology proved to be very difficult to, uh, productize and to make money out of it.

[00:05:17] So in those years, a number of companies shut down. But in the past couple of years, I would say there is a resurgence of interest and excitement about this technology in, but in slightly different ways. So now many of the companies have been shifting their focus from full self-driving to autopilot systems, similar to, for example, what you might have on a Tesla vehicle. And those systems provide value and also are actually quite profitable. So that's why a number of companies have, had been redirecting their interest toward that use case. We still have a few robot taxi companies, but now we have a lot of companies focusing on very advanced driver assistance systems. Again, the autopilot of Tesla is a good case in point. 

[00:06:09] Russ Altman: Right. 

[00:06:10] Marco Pavone: And, um, and technologically, as I said before, um, these, all these new AI techniques that have been introduced in the past couple of years have provided a boost in, um, uh, the reliability and performance of, uh, this technology. So overall, I will say that right now we are in a upward trend, but, uh, with the scope that is a bit different from the one that we discussed, uh, seven years ago.

[00:06:36] Russ Altman: Now, I know you, you're, you think very deeply about the technical challenges and I definitely want to get to that. But a social non-technical challenge to some degree is the public's perception of these autonomous vehicles. And not just safety, but like the weirdness factor of being in a car that doesn't have somebody holding a steering wheel. How, are people getting used to this? 

[00:07:01] Marco Pavone: Now, I'm not a UX expert, so I cannot really give you a quantity. 

[00:07:06] Russ Altman: Of course, this is why I ask you.

[00:07:08] Marco Pavone: But my perception is that in general people are in awe for the first ten minutes, and then they feel completely natural and they don't bother anymore about the technology in the robot taxis space.

[00:07:25] And this is actually a bit of a problem in the autopilot case because we want to keep people engaged. Um, so we want to make sure that people also understand a little bit, uh, the limitations of this technology, at least in the autopilot setting. And, um, so this is where actually engineers should have a clear conversation with the public to keep them informed. 

[00:07:46] Russ Altman: Yes.

[00:07:46] Marco Pavone: It's a very powerful technology, but still has limitations. 

[00:07:49] Russ Altman: Okay, so let's move to AI. You said, the first thing you said actually was that the, uh, and we've all seen this obviously in the last couple of years, the proliferation of AI applications and really performance in some areas that is quite remarkable. So tell me the ways in which AI is useful, uh, in the design or, uh, implementation of these systems. 

[00:08:13] Marco Pavone: There are multiple ways, of course, AI is, uh, a very, uh, large field. And for example, AI techniques for simulation are finally making simulation integral in the development process on autonomous system. But probably one of the most exciting AI advancements is represented by so called foundation models, large language models, such as ChatGPT being an example of them. And these are models that are trained at internet scale. And so you may ask, well, why does it help? 

[00:08:46] Well, the reason why a human person can learn how to drive in a few hours is that the human brings to the task of driving a lifetime of experiences. So you have accrued during your previous eighteen, twenty years a lifetime experiences in terms of detecting objects and the reason about how the world works and so on and so forth.

[00:09:08] Russ Altman: Yes, I was on a bicycle for twelve years before I touched the car. 

[00:09:12] Marco Pavone: Right, right. So basically, uh, this foundation models allows you to bring in all these generalist experience that can be learned from videos on the Internet to the task of autonomous driving, dramatically reducing the amount of data that is vehicle specific. So this provides a big boost in performance, particularly in terms of reasoning about very difficult events that if you just try to experience while driving, maybe you never see. 

[00:09:43] Russ Altman: Yes. 

[00:09:43] Marco Pavone: But if you think about internet scale, maybe there is something close enough, not maybe related to task of driving, but it still allows you to reason about those corner cases.

[00:09:52] Russ Altman: This is exactly what I wanted to ask you about, because I remember in our last discussion, I think we talked about, um, the importance of getting the edge cases. That you were training these models but you needed to see the weird events. You know, the clown on a bicycle or the baby in the carriage that's being pushed by a dog, you know, crazy things.

[00:10:11] Um, and I was wondering, as I was preparing for our discussion, if the ability of these large language models where you can tell them, like, you can, of course, you can tell them, write me a poem or write me a short story. But now you can maybe say, and this is a question, even though I'm saying it like it's true, um, create for me unusual situations that a driver might face. And I was wondering if these foundation models are any good at creating a useful set of edge cases that make your training of the drive, of the driving vehicles, uh, more robust. 

[00:10:44] Marco Pavone: They are. So probably seven years ago when you asked me this question, I didn't have a pretty good answer. Uh, this time I think I do. And so basically what these models allow you to do is two things. As you alluded to, when you design these models, you can use them to produce in simulation those corner cases. Of course, you want to make sure that you don't just generate arbitrarily crazy scenarios. 

[00:11:11] Russ Altman: Right. 

[00:11:11] Marco Pavone: But there are different ways to actually to ground this generation process in something that is plausible. For example, something that we have been doing with large language models is to mine all the police reports that have been and all the crash reports that have been collected in the United States, feed them into a large language model and use it to create a scenarios that are initialized from those crash reports. So that's basically the link to reality. 

[00:11:39] Russ Altman: Yes. 

[00:11:39] Marco Pavone: So we generally know that they're similar, maybe a little bit different, but actually close to what has happened. And then, of course, during the operation of the system again for the reasons that I mentioned before about leveraging generalist experience. This instance may allow you to reason about those corner cases, maybe by making correlations with weird things that have seen from, I don't know, the internet. 

[00:12:04] Russ Altman: So one of the things I also wanted to ask you, and I think it's related to all of this, and it may be related to AI, is one of the things that I know you publish about and you study is where the cars are autonomous, but they're also communicating with the other autonomous cars on the road. And I find that fascinating because that requires cooperation, not just between the cars and the owners of those cars. It requires the manufacturers to agree on standards for communication. And I was wondering how that's going, because I could imagine that being not only technically difficult, but also again, socially difficult. Where now the socialness is the companies that are both competing, but they have an interest, to some degree, in cooperating about standards for interaction of their products with one another. So how is that playing out? 

[00:12:55] Marco Pavone: I mean, you summarize very well the challenges related to vehicle to vehicle connectivity and vehicle to infrastructure connectivity. And this is the reason why, to a large extent, companies working both in the full self-driving car industry and in the semi-automated car industry are not relying too heavily on, uh, communication with the outside entities being either vehicles or infrastructure. At some point there is, of course, some level of connectivity. For example, you want to get some data out of your fleet at the minimum to retrain your models in order to improve them.

[00:13:34] Russ Altman: Right.

[00:13:35] Marco Pavone: But in terms of leveraging the data for real time control, most of the companies I know of are not relying on the data exactly for the challenges that you mentioned. And there are even more, like for example, cyber security challenges, cost of setting up that infrastructure, and so on and so forth. And the environment is very competitive, so that's why, um, that's another challenge, uh, against introducing additional. 

[00:13:59] Russ Altman: Yes, I mean, obviously, so the reason I asked this question is, I don't know Elon Musk personally, but I have a sense of him from the press. And I know that there are these new companies, Lucid and others that are coming up. And I'm, it might not be his first instinct to say, let's take a meeting with them and get our engineers talking about how we can, uh, and in fact, the benefits would be great because you could imagine a much safer set of, um, autonomous vehicles, if they're doing basic checking with one another. Like what's around the corner, do I have a big turn coming up all this, like the stuff that Waze tells you in a kind of a crazy way. But this could be based on actual data. So, yeah, so I guess it will be a challenge.

[00:14:40] Marco Pavone: It's an interesting case where we're solving to have this problem first. 

[00:14:43] Russ Altman: Yes, exactly. Okay. But you said it very well. Um, okay. I wanted to go, um, uh, well, first of all, is there anything else that you wanted to say about what AI is doing? Because you've now, you've given us a good idea about it's useful in the training. It's useful, maybe in the communications, we talked about this just a second ago. And it's useful in real time, um, are there other ways that AI might be surprisingly useful for autonomous vehicles? 

[00:15:11] Marco Pavone: I think these are the two main ways. So how to accelerate the design of the systems and how they improve their decision making capabilities. So basically their deployment. So I think the two most important points. 

[00:15:24] Russ Altman: Now I know that in your writings in the last couple of years, I see a lot of, uh, I saw one very interesting paper where you use the phrase risk averse planning. Uh, and, you know, forgive me if that's one of hundreds of papers you've been involved with. But I was, uh, and also a lot of papers about safety assurances, um, what is the state of our understanding of how to, um, maximize the safety of these vehicles?

[00:15:50] Marco Pavone: So, typically, the way that, uh, safety is a handle in safety critical systems is more as a constraint rather than, uh, an objective that you want to maximize. So basically you have a requirement in terms of, uh, how often you can fail. And the requirements is, uh, basically stems from considerations regarding, what happens if you fail? So what is the severity of your failure? What is the exposure of the soul of the failure? So often that failure might happen. 

[00:16:25] So you start from a requirement and then based on that requirement, you try to come up with a design on the one hand and a validation strategy on the other hand, that allow you to design a system that meet the requirement in a provable way. Here by provable, I don't mean necessarily theorem, but typically more statistical analysis. 

[00:16:46] Russ Altman: Provable was a, I've seen the word provable in your papers and it's very exciting. And the degree to which you can use that word, you know, truly would, I would guess, be extremely reassuring to both to the public and to the technical community.

[00:17:01] So that, that line, we weren't, I don't think we were talking too much about proving things at all seven years ago. And so it seems like there's been an advance. And I also, I'm also very intrigued by your comment that in the old days, there was the objective was like, get the person from A to B and the constraint was, and do it safely. And it sounds to me like there's been a little bit of a change where now it's, I want you to drive safely. And I also want you to get the person from A to B. 

[00:17:27] Marco Pavone: Yes. So basically we talk about the availability of a feature and the safety of a feature. So we want to be safe, up to a given requirement. And we want to maximize availability of the feature. So basically for how often we drive autonomously. But definitely we don't want to compromise availability with, um, we don't want to compromise safety in the quest for additional availability. 

[00:17:55] Russ Altman: This is The Future of Everything with Russ Altman. More with Marco Pavone, next.

[00:18:10] Welcome back to The Future of Everything. I'm Russ Altman and I'm speaking with Marco Pavone from Stanford University. 

[00:18:15] In the first segment, Marco gave us a great update on where self-driving cars are, how AI has really helped improve them, and what the current challenges are. In this segment, he's going to tell us about how some of the same technologies are being used to create autonomous vehicles in space. It's a very crowded field, both literally and figuratively, because there's a lot of satellites up there, and there's a lot of companies trying to create new applications for space satellites. 

[00:18:43] Marco, I know you also work, and this is incredible, on autonomous vehicles in space, or in the air. So tell me, what are you doing there, and what are the challenges there?

[00:18:54] Marco Pavone: Well, first of all, with respect to seven years ago, a lot of things have changed. Uh, including, you know, the increasingly large roles that, uh, private companies play in, uh, the space sector, SpaceX from Elon Musk. 

[00:19:09] Russ Altman: Yes.

[00:19:09] Marco Pavone: Of course being one of the most notable examples. Um, new missions, new things that we have learned, particularly regarding, uh, um, exoplanets that is, uh, planetary systems outside of the solar system.

[00:19:27] And in this context, uh, autonomy again is playing a big role. For example, there is a lot of emphasis to go back to the human, sorry, to the Moon. And of course, a prerequisite is to make sure that there is enough infrastructural capabilities to build an outpost on the Moon. And since you can't really, you know, transfer workers easily between the Earth and the Moon, it stands to reason that most of that work will be done by robots. And, for example, we have collaborations in that regard. And, again, AI is going, is poised to play a big role in the design of autonomous systems for these new space applications. With understanding that, of course, the environment is different.

[00:20:17] Russ Altman: Yeah. 

[00:20:18] Marco Pavone: Spacecraft are very limited from a computational standpoint in terms of, uh, amounts of data that came like crunch with respect to what you can do on a self-driving car. And that is very limited, like to a large extent, you can relatively easy gather data, uh, for self-driving cars by just driving around. But gathering data in terms of how to land on the Moon, or maybe on some, you know, satellite of Jupiter, you don't have the data. 

[00:20:50] Russ Altman: Right.

[00:20:50] Marco Pavone: And so in a context where data is becoming increasingly important, how do you devise techniques that are much more judicious in how you use that data? So, there are a lot of promising studies, there are path forwards, but, uh, overall, the strategy will be different. 

[00:21:08] Russ Altman: Are we mostly thinking about unmanned, unpersoned vehicles or are, because that takes away a whole bunch of safety. I'm sure you still don't want these things to blow up and crash because they're expensive, but it's different when there are humans involved. So is human, uh, part of the equation right now, or is that, is it mostly entirely robotic systems?

[00:21:29] Marco Pavone: I mean, in some cases, there is a debate whether you want to have a human or not. In some other cases, there is no debate at all. So if you want to explore, uh, icy bodies in the solar system, that for me are among the most exciting targets for space exploration, such as, for example, Europa, around Jupiter or Enceladus. 

[00:21:49] Russ Altman: Yup.

[00:21:49] Marco Pavone: You can't send a human there. 

[00:21:51] Russ Altman: Right, right. 

[00:21:51] Marco Pavone: It takes literally too long. And maybe the human will be able to get there, but definitely wouldn't have time to come back. 

[00:21:58] Russ Altman: So nice and easy. 

[00:21:59] Marco Pavone: So there you want to have a robotic, uh, fully manned platform. For other tasks, uh, like, for example, going back to the Moon, you probably want to have a combination of astronauts and also robots that take care of the, uh, dull jobs, for example, construction jobs. And then for operations around the orbit, like, for example, for in orbit assembly or maintenance of space assets, again, there is a strong push toward robotic systems from, you know, an economic standpoint. 

[00:22:39] Russ Altman: Yes. 

[00:22:39] Marco Pavone: To basically minimize cost. So the short answer is, yes, there is a strong push toward increasingly more amended platforms.

[00:22:47] Russ Altman: Are there similar issues for as for self-driving cars in terms of coordination of multiple vehicles or can, or is space so big that you can make the assumption that it's a freedom to operate and you don't really have to worry about other devices in this space?

[00:23:05] Marco Pavone: Well, believe it or not, collision avoidance is also a problem in space.

[00:23:09] Russ Altman: Okay. Okay. 

[00:23:10] Marco Pavone: With all the junk that you have around the Earth orbit. 

[00:23:14] Russ Altman: Yeah. I think I recently read that Starlink either has or will have six thousand satellites and is working towards thirty thousand. Does that sound? 

[00:23:23] Marco Pavone: Yes. Yes. That's right. And especially in the past, people were not very careful in, uh, decommissioning space assets. So that's all the junk that we still have. 

[00:23:35] Russ Altman: And that junk does not communicate. Does it even put out a signal to say I'm here?

[00:23:39] Marco Pavone: I mean, you can see it through sensors. 

[00:23:41] Russ Altman: Okay.

[00:23:42] Marco Pavone: But it doesn't communicate or definitely doesn't cooperate in order to. That said, in terms of communication, actually that plays a big role in, um, in a space missions in the sense that, uh, the idea of sending, uh, groups of satellites or space assets that jointly achieve a task that will be impossible or very expensive with a monolithic mission is actually becoming almost the norm.

[00:24:14] Russ Altman: Is this because like the thrust required to get them up into space is much cheaper if you break it up? 

[00:24:22] Marco Pavone: Part of it, uh, and the other is that, of course, by distributing your space assets. For example, you might have a much better coverage than what you could achieve with a monolithic architecture. Or, if you want to do it with monolithic architecture, it actually might need to require much more expensive sensors.

[00:24:39] Russ Altman: I could also imagine, you were saying before that we, it's hard to get training data for landing on the Moon or landing on Mars. And I could, and I'm making this up and this is not what I do, but I could imagine that if you have a bunch of vehicles headed towards the Moon, there's a certain triangulation they could do just with each other and with various, you know, stars and other, you know, milestones to perhaps do more accurate estimation of the flight. Is that a thing? 

[00:25:06] Marco Pavone: Yeah, absolutely. That's one of the ways that distributed space systems might actually work well by exploiting collaboration. And if I may advertise a new research center we have set up at Stanford, together with a colleague of mine, Professor Simone D'Amico, we have started a new center called the CAESAR.

[00:25:26] We're both Italians. So, there is a, there is a legacy in the name. And this center is all about, uh, investigating ways to use AI techniques judiciously in the design of, uh, distributed autonomous systems for, uh, space exploration. 

[00:25:46] Russ Altman: Okay. So that's very exciting.

[00:25:47] Marco Pavone: Got more to tell you in seven years from now. 

[00:25:50] Russ Altman: Yes. So, and actually I want to go back because you led off this discussion of space by saying it's very different because of private industry. And so tell me a little bit about that, because this is kind of a new thing. In the old days aerospace meant NASA, right? That was like, I grew up at a time where they were synonymous. Um, what changes from a technical point of view are introduced when you have multiple players? We already talked about on the road, the difficulty of communicating with a competitor's car. Is it any better when it comes to the space race? 

[00:26:25] Marco Pavone: I think there are a number of reasons why, um, private entities have, uh, become involved in the space sector. One of them is technological innovations. For example, in the past twenty years, we've seen how you actually can do excellent science or uh, fulfill, uh, useful tasks with miniaturized space assets.

[00:26:54] So you don't need, uh, um, a satellite that is as big as a container that by being very expensive, but even a shoebox type of satellite can already provide the value. So that of course, has, uh, open up opportunities for a number of stakeholders that, of course cannot invest billions of dollars in space missions. So there is one aspect of it. 

[00:27:16] The other aspect of it is that a number of opportunities have opened up for business. For example, using space assets for communication, for surveillance, for logistics. And the people have been very quick in identifying those opportunities. And together with the technology innovations that now make those opportunities, uh, profitable. That's why you have several, uh, private stakeholders out there now. 

[00:27:45] Russ Altman: This is very exciting because I just, the list of applications you mentioned were, went way beyond what I was aware of. And I can see that really almost every industry can now think about what a little shoebox flying up above the Earth be useful to my logistics or to my business. And the answer might be yes, a lot of the time. And this is also obviously an international challenge. How is the level of communicate, of collaboration across all the different international players? 

[00:28:14] Marco Pavone: Yeah, that is a bit more, a little bit outside of my domain of expertise. Of course, space is, uh, a topic that unites mankind, but at the same time is a very sensitive topic from a military standpoint.

[00:28:31] Russ Altman: Right. 

[00:28:31] Marco Pavone: So with some countries, of course, there is a close collaboration with others. As you can imagine, right now, there is a much more tense, uh, conversation. 

[00:28:40] Russ Altman: Thanks to Marco Pavone. That was the future of autonomous vehicles. Thanks for tuning into this episode. You know, we have more than 250 episodes in the archives, so you can find pretty good conversations about the future of almost anything.