Beep’s AI-Enabled Transportation Solutions: Connecting Communities and More | InTechnology | Intel
(buoyant music) - You are watching, "In Technology." This episode is one in a series with innovative companies who are part of the Intel Capital portfolio. In these conversations we'll explore the key areas of technology Intel Capital invests in to shape the future of compute. Hi, I'm Camille Morhardt, host of "In Technology," and today's episode is part of the Intel Capital series.
We're going to cover autonomous transportation with a company who is building a service. Mark Rostick is co-hosting with me. He's Vice President and Senior Managing Director of Intel Capital. He's co-manager of Intel Capital's Cloud domain and a voting member of its investment committee.
He's responsible for approving investor proposals. His focus is on companies that innovate with machine learning and artificial intelligence as well as enterprise services that manage cloud infrastructure. Welcome to the podcast, Mark, and who have you brought to the table today? - Hey Camille, thanks. Today we have two executives from one of our portfolio companies.
We have Joe Moye, the CEO, and Kevin Reid who's the chairman of the board. - So tell us a little bit, Mark, before we get into the conversation, help us set the context for what Beep is doing and what the industry is and why what they're doing matters right now. - Just by way of background, Intel Capital obviously makes investments on behalf of Intel Corporation for two reasons, one is to earn a good return on those investments, but more importantly, we make investments in companies that we think will help Intel learn, advance its products in the market or understand perhaps new technology trends that are coming. And we made an investment in Beep for a lot of those reasons, A, we've known the team at Beep for a long time, we've worked with them in the past. Intel has been an investor in autonomous transportation for a long time. Intel obviously owns Mobileye, which is a pioneer in the space.
And what Beep was doing to try to bring autonomous-based transportation to underserved markets was something we felt was really compelling. Not just because new markets were being reached, but because they were doing it in a way that dovetailed really nicely with where we thought autonomous transportation and the technology being used to do it should go. - Joe, tell us, what does Beep do? (Camille laughing) - Beep was founded to really solve for primarily the micro transit deserts that exist in our society and specifically in the US market, to provide and extend mobility and accessibility to our citizens is a really important gap that we aim to solve and we're solving that through the use of autonomous, shared, so a form factor of a 10 to 15 person vehicle to start with, electric platforms that are ultimately orchestrated to provide these mobility services in urban areas and campus environments, those airport type use cases where you've got the need to move people across these, again, first mile, last mile type scenarios. So we have brought to market a compliment of technology that enables and orchestrates a lot of that service requirement as well as some of the service provisioning itself to ensure both the safe and effective use of those types of platforms in these environments.
- I guess the most basic question I have is why make those autonomous? - First and foremost, there is a significant lack of drivers out there in the workforce today which has caused us to actually reduce public transit services in many areas across our country. So this isn't a question of labor arbitrage, this is a question of bringing these technologies to play that ultimately allow for us to extend and grow the services out there in these various communities and use cases. So the autonomous technologies are a critical component of that, needless to say.
I think it's probably important to point out to your users, although ultimately there isn't a driver on board, an important piece of what we provide is retaining that human in the loop so that you're able to communicate with passengers in the event that there's some need to interface in the cabin, the ability to monitor the vehicles themselves for continuity of service but also for the safety of the passengers on board. Bringing these technologies to play affords us a chance to grow the mobility options across our country, which is an important objective of the company. - I'm wondering, Kevin, if you could give us just a landscape of the regulatory environment in this space also? - Any vehicle that goes to market has to have a certain level of what's referred to as FMVSS, or Federal Motor Vehicle Safety Standard, threshold has to be met so that the passengers are safe. And then separately from that you would typically, based on the state that you live in, one around the vehicle and the protection of the occupants and the other that is really based on the driver.
And what is very interesting about autonomous vehicles is that certain states have made critical, I would say, regulatory or policy changes or introduce different capabilities that allow for the driver to be autonomous, but we still need to make sure that the vehicles themselves are FMVSS capable. And so to Joe's point earlier on the form factor, one of critical things we've been looking at is for public transit, how do we get a large enough vehicle that we can have multiple passengers on board so we can ultimately impact congestion by removing a lot of the smaller vehicles out in the world, but at the same time have something that is small enough to be able to get to these extensions and yet abide by a set of regulations that were created when there had to be a driver on board. So things like do you need a rear view mirror, do you need a side mirror? And in the case of autonomous vehicles, of course when there's no human driver, many of those things are different. So there are a number of steps that we've been taking over the years to really address this, but it is a very interesting area, as you can probably imagine, the use cases are different.
- There's been a lot of obviously press written, controversy, around companies like Waymo and Cruise and robotaxis, not just from a regulatory environment, just what impact do they have on the municipalities they're operating in. Can you talk a little bit about why Beep is different? Maybe why is it better than the robotaxi solution at least in the next several years? - From a drivability standpoint, the autonomous driving systems are very sophisticated. So obviously by leveraging people like Mobileye, we tap into the same type of capabilities and resources that are available from companies that have spent billions of dollars trying to make sure that the vehicle can operate independently as well as it possibly can. I think if you then augment, much to Joe's earlier point, having humans involved from a supervisory standpoint gives us an additional capability of oversight that we thought was very critical for public transit. Now if you put that aside and look at number of vehicles on the road from a robotaxi standpoint, if you think about the demographics of trying to, let's say with these services like Uber and Lyft that are really more geared towards a higher price point to different type of demographic, what we wanted to do would be is make sure that we could expand mobility access to as many people as possible.
So whereas many of the robotaxi companies have gone for either two or four seater type configurations that they plan to roll out over time, if you think about Uber and Lyft as examples, they don't really minimize congestion within a city, because you still have a number of vehicles that are sitting around waiting to be deployed for one or two passengers that would essentially be trying to get from point A to point B. In public transit what we're really trying to do is get as many participants as possible into the same vehicle from a shared form factor perspective and so just the sheer number of vehicles gets minimized. And so while robotaxi also, because of their electric nature, minimize the amount of emissions that go from a climate perspective are also good like our vehicles, the difference is you still need many of them and therefore from a traffic and congestion standpoint, that really has a very different impact. - Another really important distinction is that we're operating in geo-fenced areas, managed routes, and controlled speeds. And so in that environment it's naturally gonna be more performant, more safe, and so I would argue that with some of the challenges that we've read about in the papers trying to solve for the most complex scenarios, any to any, "pick me up in my driveway and take me anywhere I want to go," very different than providing managed services in a specific geo-fenced area with planned routes, stops, and ultimate fully deployed services.
- So is your expectation that the rollout of autonomous vehicles will begin in these kind of confined geographic locations with the first adopters being like cities or universities and the fleet deployed on these specific routes? - Yeah, Camille, I would say two things to that. One is these types of use cases are being deployed in pilot forms today, both in the example of what we do as a company in these more planned routes versus what robotaxi does in an any to any scenario. I would argue that this alternative means of transportation in these controlled environments is gonna be adopted and scaled more rapidly and more early than the more complex use cases as we've described in this example.
So if we think about the impact that we wanna have in the area of these micro transit services in particular, we see in the next couple of years a real opportunity to integrate these platforms into mobility networks and public transit as we know it today. And the ability to integrate autonomous platforms into public transit isn't turning on a vehicle and an ADF and letting it go, you have to integrate into all the legacy environments. How do you coordinate with bus services in a centralized downtown area to provide these additional forms of transit or arterial routes? How do you provide capabilities like both on demand and scheduled services? How do you address the realities that you need to be able to communicate with passengers if there is an issue with the system or an issue with the route? You can't just let the passengers inside figure out on their own, "Do I get out of the vehicle? Do I wait for the next one?" You need to immediately communicate and provide some action. As we look at removing a driver from the equation, the driver does a lot more than just steer, throttle and brake, they have to be able to provide these types of services that ultimately we do both through the use of technologies and through the use of command center services that we provide. - What was something that kind of surprised you to learn during those pilots? - I'm sitting here in Lake Nona, Florida where we've been operating for almost five years now, providing a service across a master plan community development here in the Orlando area and providing use cases to medical centers, to business centers, to the residential areas, the town center areas. So providing kind of connectivity across seven different routes.
We've deployed seven of these similar type use cases in Florida, including working with the likes of the Jacksonville Transportation Authority and integrating into public transit environments, Lynx, which is the public transit authority here in Orlando in a downtown area. We just deployed in the Honolulu airport to provide one tarmac services between planes and baggage collection areas. - Any surprises? - You weren't sure when you launched your first projects, is this gonna be a novelty where people check the box, say, "I rode an autonomous vehicle," or are they really gonna adopt it as an alternative means of transportation and in all of the use cases we've served from national parks to planned communities to downtown urban core areas, it really has proven out to be something that will change the way we ultimately move people when we're able to scale these platforms much more broadly than the pilots of today.
I would also just say from a lessons learned perspective, community engagement is absolutely critical to the success of these platforms, long term. We've done things like first responder training where we bring in police fire and rescue, to acclimate them to the vehicle and the service, 'cause, it's an education at all levels, people don't think about it, but if you just throw autonomous vehicles on the roadways and they come into a scenario where they need to interface with our first responders, how do you make sure that they're familiar with it, comfortable with it and can interface with it in a way if they need to provide some level of support that's able to be done? Those are a couple of the things I'd hit on. - Joe, if somebody asks you, is Beep a transportation company? Is it a technology company? How do you characterize yourselves and how do you differentiate between them? - We are a technology company, we are technologists at heart.
That being said, as we were working through our business thesis and entering the autonomous mobility space, for all the reasons we've described, we settled in on the need to get into these more controlled use cases, the impactful micro transit shared mobility and, yes, electric platforms, because of the compounding effect on carbon emissions. But at the end of the day, as we learned more about transit, public transportation, and delivering that service, we quickly identified that was the gap that we needed to fill both from a technology perspective and a technology-enabled services perspective. A lot of folks entered this market thinking we just need a vehicle and an autonomous drive system and we're good to go, launch a service. And as we've tried to describe, there's a lot more to closing the loop to provide a great quality rider experience, a safe experience, and one that people are ultimately gonna utilize. And a lot of that's been driven by the technology that we develop and deploy, but also a lot of that has been an understanding of how transportation services need to be provisioned and how you need to augment beyond just the vehicle and the ADS.
- As technologists, we had to look at it as, first of all, how do you create a solution that works for the people that use it? So if you're gonna have something that's public transit, how do you make it appealing, but also safe? We've learned for instance, that you can't just think about even the passenger experience inside the vehicle, but what happens prior? What are you doing to actually help communicate from a public transit standpoint when the vehicle is available next? Or what different types of services might be available based on where you go? And so that whole concept of the experience even translates to other road users. If you're at a pedestrian crossing and this vehicle comes rolling up, the natural human tendency is, you look the driver in the eye, make sure that they see you and then you cross and in the absence of a driver, what do you do? So we've actually had to work with the automotive companies as we've gotten that experience to take back and say, hey, we have to think about other road users as well and figuring out how you can leverage tools like AI and historical information, because we partner with the public transit organizations, we have access to data, as part of the community engagement and part of the planning, we also look at what routes make the most sense based upon where people are typically trying to get to and also how do you recycle the assets, because one of the benefits of autonomous vehicles is you could task them and you could task them multiple times throughout the course of a day. So you could have three vehicles running a particular route from seven to eight o'clock in the morning if you're trying to move essential workers and then redeploy those vehicles to other areas after eight o'clock if that particular route doesn't necessitate that sort of capacity at some other point in time. And so, you can really start to move around the assets in a much more effective way, because you're looking at optimizing for that community as opposed to optimizing for a particular rider. - I think people are accustomed now to the idea, the notion, of riding in an autonomous vehicle, but I think everybody generally accepts that while everything isn't perfect yet, machines don't need coffee, they don't get tired, they don't drive drunk.
So like we know that in theory, especially on pre-planned routes, this could be fairly safe. I think what's pivoted in fear in people is can it be hacked and what happens if it gets hacked and how do we protect? Can you guys talk a little bit about cybersecurity and what you're thinking on those lines? - Obviously at the basic level, adhering to the the ISO 20, that 7,000 series at a vehicle specific level, the ISO and SA, the the functional safety two requirements, all of that is in place as part of this entire stack. Those protection layers are preventing and or identifying intrusion, on top of that is part of what we integrate into the vehicle itself is our own independent auditor. We know where a vehicle should be at any point in time if it's provisioning the service that it was designed to operate within. And so, if it deviates from that, we have the ability to immediately identify that and we actually have the ability to enact an emergency stop if there were something that we felt was unsafe or if there was some other event that may have happened. So I think that's a really important point and a really important distinction.
And the last thing I would say is how we integrate with the autonomous drive system in our remote command and control environment. We are not believers in tele-operating vehicles, we rely on the ADS itself to ultimately provide that maneuvering and decisioning. - ADS, autonomous driving system. - Yeah, correct, the autonomous driving system and stack. And you think about the difference of tele-operations where you can literally connect into the vehicle and take over control from the ADS versus what we call tele-assist, where we can help the ADS make a decision because it does not have the information it needs, in my example of pulling off safely and letting passengers exit at a proper point in time, because of some issue be that roadway or technology related, the thing I would leave with your listeners is not only are all the standards in place to ensure kind of safety from external intrusion, but also extra layers of safety in the services that we provide to ensure that we've got protections in place to optimize how we protect those riders.
- That independent auditing is really important too, because if you think of it, it would be very difficult for hackers to hack multiple systems at the same time and take control. And so if you think about the concept of the vehicle capable of making its own decisions about what's safe or unsafe and then a completely independent group as an operator, like me, being able to have tasked the vehicle with the mission, knowing where the vehicle should be, independently from their system, verifying that the vehicle is where it should be and is doing what it should be doing and if not being able to kill it. But if we were hacked for whatever reason, again, you have all the best protection mechanisms in the world, if we happened to be hacked and a mission was given to the vehicle, wouldn't perform anything that is unsafe either. There are all these backups, even to the point of physical sensors, If you take off a device, for instance, Mobileye, as an autonomous drive system has fully redundant paths that say if you somehow had a mechanism where you're taking over from a camera perspective, I have an independent path with radar and lidar and so, there's so many different layers for that physical layer and the software layer that you can imagine having that concept of defensive depth gives you enough layers that one particular intrusion could not lead to the breakdown of the overall system. - Is there also like an analog shutoff capacity in the vehicle itself? If there's passengers who notice something or whatever you can't predict, is there a takeover or a shut down? - Not a takeover in the sense that we don't want customers to be able to take over control over the vehicle, but there is an emergency stop capability in the event that someone believes something really bad is happening that can be done physically as well.
And these are many of the things we've had to work through over the years and so you also prevent that from being a safety issue with someone hitting that the middle of a busy street where there are cars also behind you and not having an unsafe incident because of someone doing something as well. So there are so many aspects to this that we've had to work through with our partners of course because we don't make the vehicles ourselves, but it's all in the course of public interest. - This is a lot. You have to be good at a lot of things and so it's hard to build a company like this. Why is it worth it? Why is this opportunity so big? What do you see as the prize here for getting this right? - I have to say this is the first business I've been part of, a venture-backed business, that yes is a technology play in a lot of ways, but it has such a broad societal impact. If you think about improving the quality of life of our seniors, who 25% don't have access outside of the homes and being able to extend their mobility.
If you think about the disabled in our society and an opportunity to really provide new access to goods and services, being able to now extend services into underserved areas of our downtown urban arenas to bring people into jobs, healthcare services, it's really the most impactful technology that I've ever been part of that really has a chance to change a lot of important elements of our society. - The idea for Beep came from my dad who was an engineer, he got his engineer degree from Howard University, but while he was pursuing that it was a very different time back then, but he really struggled to get to school on a regular basis and he was fortunate to navigate through that, but he ended up having a lifetime passion around trying to make sure that he could make a difference in terms of mobility access for all. And so he went into fleet management, he was a fleet manager for the city of Gainesville when he retired.
And one of the things that he realized as he retired was that there were a lot of consulting opportunities looking at the intersection of what Joe mentioned earlier about shortage of drivers in public transit and the role of electric vehicles, as electrification came in, and everyone was concerned about climate issues. And so, a lot of what we came up with here was really an idea that came from him and his passion around how do you extend this mobility access to all. And he passed during Covid sadly, but he was part of the initial genesis of this and I would say probably got through a good two, two and a half years of our early foundings on this. So he saw the direction we were going in and shaped a lot of what Joe and I ultimately came to think of as the strategy for Beep.
So it's great to be able to do good for the community and at the same time have commercial traction like we're having in an area where we can execute well then we can also make good money for venture backers as well. - Thank you for explaining that. I don't wanna let you guys go without understanding a little bit, even if it's quickly, about how you use AI in training the models and how you incorporate realtime or near realtime data into the driving decisions. - And so much of what we've done over the years is leverage AI models to figure out what types of circumstances create incidents for us. And we found a lot of things that surprised us, just as an example, vegetation, as we started to train these models and understand where we were having significant issues and why and looking at the trends, we actually found that in many cases a lot of the issues we were running into involved false positives were around vegetation where trees would grow into an area and it would confuse the autonomous driving systems, because yesterday when it ran by there wasn't something there and today it's there and what's the problem, right? So from a safety standpoint, a lot of what we're trying to do is train the information so that we can build better missions and tasks. So in other words, we don't own the autonomous driving system, but what we do is to learn from what the issues have been in the past and as we determine that a route needs to get from point A to point B, we take that learning to build the safest path so that the vehicle can get from point A to point B in the best way possible for the autonomous vehicle operating domains, that's the first piece.
Then from that also, we interact with the infrastructure and that's not really AI, that's just straight up integration with cameras, traffic systems, et cetera. And the goal here is to leverage AI with computer vision in those cases to see if we could do early detection of a problem, let's say, in an intersection that a vehicle is approaching. But what we do is we then send alerts to the actual autonomous driving system, for many different reasons not least of which could be external hacking or other purposes like that. We want the autonomous vehicle to make its own decisions. And so you take the real time sense, for us is more the alerting and think of it, if you were a driver and someone were to say, "Hey, someone just told me there's deer around this corner," then maybe I'll slow down, I'm not gonna necessarily slam on my brakes immediately, I'm not gonna take an action other than set myself up better from a defensive standpoint to make sure I can make that right decision. So it's really more information provisioning that we're doing on the real time sense, but the AI for us is much more about how do you mine and then how do you use computer vision to really drive the identification of scenarios that may not be safe.
- I would just add one other quick example of how we're using AI technologies onboard. Again, picture the scenario in public transit where you don't have a person of authority, a bus driver or shuttle driver onboard the vehicle. How do you identify a health event? Somebody falls onto the floor or is exhibiting signs of having a heart attack? How do you identify somebody has taken out a weapon or there's some other uncivil event where you need to intervene with the vehicle.
The use of those technologies as predictive technologies to ultimately be able to identify scenarios that alert our command centers and ultimately enable us to integrate into the vehicle virtually dispatch first responders, whatever that action is, that's an important piece of that equation for safety and ultimately gaining ridership and comfort with these vehicles over the long haul. - Kevin Reid and Joe Moye, Chairman of the Board and CEO of Beep, autonomous transportation company, thank you for joining Mark and I on this episode of "In Technology." (buoyant music) - Thanks guys.
- [Announcer 1] Never miss an episode of "In Technology" by following us here on YouTube or wherever you get your audio podcasts. - [Announcer 2] The views and opinions expressed are those of the guests and author and do not necessarily reflect the official policy or position of Intel Corporation.
2024-08-29 07:21
