AI and Robotics in Robotaxis

Added: 2026-06-04

[00:00:01]A day of history and hope.

[00:00:07]>> Asalam alaikumah and welcome to open forum from MTA International USA studios. A space where ideas meet dialogue and questions of today shape the answers for tomorrow. In each episode, we attempt to take on the most pressing, fascinating, and perhaps most challenging issues of our time. We attempt to distill complex ideas into simpler forms of understanding and creating dialogue for the future. So let's begin. Today we will be talking about cars that drive without anyone in the driver's seat. Those are called robo taxis. And there are companies like Whimo and Zuks and others that are attempting to make strides in this area.

[00:00:56]And they actually exist in some cities around the world like San Francisco, Phoenix, San Jose, and a couple of other areas where they're actually driving these cars without a driver in the seat.

[00:01:08]And they have millions of miles of data to interpret. To go into this technology and understand what it means for neighborhoods around the world, we have with us today Amar Ahmed Sahib. Sayya demar Ahmed Saiib currently works at Ford in the connected and automated vehicle space. He is a has a PhD from Virginia Tech in electrical engineering.

[00:01:32]Now Amarab Jazaka for being here with us.

[00:01:35]>> Thank you. Let us now take a minute to do a deeper dive and break away for a second and we'll come back.

[00:01:42]Imagine stepping into a car with no steering wheel, no pedals, no driver's seat, just two seats facing each other. That is what we would call a robo taxi.

[00:01:56]Companies like Whimo and Zuks have been building driverless cars from day one, built with absolutely no driver in mind.

[00:02:07]It is as if this car has superhuman reflexes predicting what exactly might happen.

[00:02:15]But there are challenges. If sensors get blocked by snow, mud, or glare, the vision could get worse. If maps are outdated, the car might get confused, and people may not trust a car with no human at the wheel. So the real question isn't can cars drive themselves rather it is can humans relax enough to let the car drive itself. Today we'll talk about trust and comfort and what happens when the wheel disappears as we have examples of robo taxis around us already.

[00:02:58]>> Jakamala for that brief and concise explanation. So robo taxis or cars without drivers where I can sit in the back seat and it will take me from point A to point B exists. I've actually ridden in one. It's very fascinating and a little scary to might I add. Um what let's start with the passenger experience.

[00:03:19]Should I be riding in one or is it too early and very unsafe?

[00:03:23]>> We can say that at this point the technology is developing. It has come along far away far far uh it has traveled basically it has gone and developed over the past many years from its infancy stage where people were thinking about the idea and now we are at a point where people are beginning to drive in prototypes and other proof of concept >> I remember when there was adaptive cruise control where it would from the car in front of me the car would slow down and speed up now we don't even have someone in the driver's seat so it has evolved the technologies come a long way >> indeed it has and I think what has happened is that a lot of things that were basically being developed over the years have come to fruition at this point. Um the key point is that uh the focus has been to moving towards what's called level four autonomy which is the complete absence of a human driver and basically the passengers do not have to uh be you know be in control of the vehicle at all. And so these >> what's that experience like? I mean, if I get into these, I I believe they're Jaguars with Whimo and Zuks um um and others are doing this as well. What is the experience like if you get in?

[00:04:29]>> Basically, you get in and this is going to introduce the vehicle would introduce, you know, welcome you inside and then you will >> you use a phone to book it like an Uber or a Lift vehicle.

[00:04:40]>> Indeed, uh you can uh you know, book it through Uber or other apps similar to that. Uh once you get in, it will welcome you. Basically, it will tell you, you know, you're going to go from point A to B and then it will basically try to make yourself comfortable as much as it can and then um as it goes along, you know, it's traveling like a regular vehicle. It's just that uh there is no driver present and that's what's called level four autonomy and it's using its own onboard sensors to basically make the decision. So, there's a little bit of uh you know novelty there because you do not see the driver. It's all automated.

[00:05:13]>> It's a little scary >> indeed. It can be. It can definitely be u you know counterintuitive to move around without a human driver in there.

[00:05:21]But the difference is that if you're looking at a more conventional vehicle like for example Veimo, you might see the steering move around but there's no driver. It's the the ghost is driving in the autopilot itself. Uh in contrast to that you have uh Zuks which is more of a shuttle and gives you an experience similar to what you may experience in a train and that's a little more comfortable because we kind of have the hang of that experience where there's no driver but we have kind of driven around >> but but how does how is this technology coming so far? I assume there's cameras.

[00:05:53]I've heard of LAR and wireless sensors that can tell around. Um, but what about the human experience? If there's if there is construction and they figured all these things out.

[00:06:05]>> Well, that's where the challenge lies because on a sunny day, all these sensors can detect threats. They they follow the roads very well. And so the the fundamental challenge that this technology has had to cover are edge cases. So for example, what if the sun is right in front of the vehicle, the cameras, would they be able to detect that or not? or what if someone is walking past the vehicle, you know, a pedestrian, a child. That's where um the technology has evolved and come leaps and bounds over the past many years, but still there are many edge cases and challenges. And that's why some of these companies are focused more on control environments where they'll know the area, the the weather is nice, they don't have to deal with fog or other challenging issues. Um that's where they're trying to still get more data so that they can move along and make the incremental progress that's needed to come to a point where it can be scaled up to other cities.

[00:06:54]>> And what do they do because they're doing in few cities they map that city every corner, every stop sign, every curb u very closely so they understand the environment around them. Uh is that what's happening right now or they're just trying it in a few cities?

[00:07:10]>> Well, they're trying in a few cities. Um they obviously had to have uh permission from the local government from the state to have that uh autonomy to drive which means that they have some right of data they have right of you know knowing information but the biggest challenge or the solution that they're trying to aim for is uh the fact that for a large city um to know every nook and corner is going to be a challenge. So therefore they want to try it out in certain locations gather the data and use those learnings to scale it up to other locations. So that's where I think the focus is not to go you know pretty fast but rather to understand the challenges >> and what about the computing power required I from what I understand there's hundreds of millions of decisions being made every minute because there could be people around you there could be someone crossing the street or they may be crossing a street it needs to judge that it needs to find where the light is how does all this come to there's a massive amount of intelligence um that's there but it's not as strong as a human's ability to assess those things.

[00:08:09]>> Well, human beings have this cognitive ability. The machines are learning based on human experiences. So, what may have happened is humans have driven around the vehicle. They have gathered the data and then use that to train their models.

[00:08:21]Now, the issue is that um that model is based on prior human experience. So, those machines will not have information about new and novel challenges or you know situations that may arise. So that's where um the sensors are designed to operate in certain ways and they would only be able to follow prior experience.

[00:08:40]>> And and now um what is the technology inside the card? I assume there's lots of cameras and there's a a chip a GPU that captures this information is interpreting it.

[00:08:53]>> Well, you can look at multiple sensors that are part of the vehicle. How does it all work?

[00:08:57]>> Sensors like cameras are all around the vehicle. Well, you have cameras that can detect objects and pedestrians. Then you have GPS which gives you your location.

[00:09:05]Um, you may have other types of sensors such as radars or lighters which give you exact ranges.

[00:09:10]>> What is LAR?

[00:09:11]>> So lighter is the laser equivalent of a radar. What it means is I can fire a beam at a certain target a human being a vehicle and I know exactly what the distance is. Now in contra >> if you fired a beam at this mug you would get the distance. I would know this is about you know uh you know 3 ft from me roughly give or take.

[00:09:30]>> Let's pause for a second and bring in another expert in this space. We're going to have joining us is Naser Bachce Sahib from Silicon Valley who is a PhD in AI as well. Nasar Sab thank you for joining us.

[00:09:48]>> Asalam alaikum. Uh thank you for having me. Uh now Suzab share your background in technology and you have a PhD in this space from University of Tennessee Chattanooga. You're originally Ganan. Um what um what's your background in technology?

[00:10:05]>> Um I graduated with a PhD in computational engineering from University of Tennessee at Chattanooga.

[00:10:12]I've worked in robotics for about um eight years now in different roles and currently I'm I'm going deeper into the platform space uh for the infrastructure that backs um AI.

[00:10:26]>> So to today we're talking about autonomous vehicles and from a robotic standpoint as well how do these machines robots speak to each other? How do they make decisions? How does this work?

[00:10:41]Okay. So you have to think of it um like um these machines need to be able to solve problems similar to or at the level of um human operators. So for example, if you are thinking of a self-driving vehicle, think of it like a robot that is on wheels. It first needs to be able to see its environment.

[00:11:03]Seeing is in two ways. One is to be able to observe the ambient light to be able to know what is there. But in rare conditions, hazardous conditions like in the night where light, natural light isn't bright, it needs other tools like radar, like lighter to be able to allow it to be able to recognize everything around it.

[00:11:26]>> So this is back to these signals and communication to be able to see around LAR. This is just wireless communication and signals.

[00:11:36]>> Um, mainly wireless because because it's autonomous, it cannot be plugged to the wall when it's operating, >> right?

[00:11:44]>> Meaning that it needs to be able to carry all the tools it needs with it on board and need to be able to use them without connecting to anything um to guide it.

[00:11:55]>> So, how does this work? Uh, once it's not connected to anything, uh, how does it communicate? H how do you make it work?

[00:12:04]>> Um the first thing is that it needs to be able to bring all the information together in a process called fusion. So fusion allows it to be able to bring all of the environmental signals including 360° pictures into um one logical unit where it is able to make the decisions.

[00:12:24]the decisions include um where to navigate and what actions to be able to take.

[00:12:30]>> That that that's really helpful. How how about we pause for a second and switch to another topic. We had um the privilege of uh and a chance to take a ride in the Whimo which is Google's uh technology in San Francisco, California in the USA alongside with some guests visiting um USA for for a conference and we were able to share that experience and we have some video here. Uh let's take a look at this and see what the experience looks like. During the recent AMDIA conference of entrepreneurs hosted by the Senat department where technology and AI took center stage, we operationalized the conversation by taking it directly to the streets of San Francisco, California. Our MTA USA team conducted a firsthand field test of a Whimo autonomous ride showcasing the realworld impact of emerging mobility technologies.

[00:13:33]We were also joined by a distinguished attendee, the managing director of MTA International, Monito Shamsahib.

[00:13:42]>> Hello from Whimo. As we get going, just give us one minute to cover a few ride tips. This experience may feel futuristic, but the need to buckle up is the same as always.

[00:13:59]If you're traveling with little, >> welcome back to the studio. Um, so now that you've seen the Whimo and and some guests experiencing, I had a fascinating time and I've been inside uh Tesla, which is not as advanced. You still need a driver and Whimo, it's a night and day difference. Let's talk about Zuks now.

[00:14:28]What's this birectional seating? And they have a very different experience that they're going after.

[00:14:33]>> Indeed. I think the objective from Zuk's point of view, of course, this is one of those companies which is controlled by Amazon. They're trying to attract attack a different niche market which is that they will scale up uh these um robo taxis which are more of a cabin feeling for for a train and where you do not have the regular you know uh driver vehicle experience that people typically associate with a vehicle. This is more of a train experience on a public roadway which you might see in San Francisco for example. And so they're trying to attack the problem for a from a very different point of view which is that give a very different perspective and human experience. more of a human psychology issue where I feel uncomfortable watching a steering wheel turn without a human there so put me in the back of the bus so I don't have to feel that >> right that is correct so these are more of a fleet vehicle operations where you have dozens of vehicles operating around a city and they can provide you this customized experience where you have control over where you're going where which routes you can you can possibly select from and they would for example zuks would not I think uh you know go on highways beyond a certain you distance or region. So that's why the experience is very customized and using the learnings the company would aim to scale.

[00:15:45]>> This is kind of like from point A to uh to the airport and back something like that mirror uh use cases. Interesting.

[00:15:52]Very interesting. You know what? We're going to take a pause. We're going to go to the streets with our street reporters and hear directly from the American public how they're feeling about this.

[00:16:02]We are here in the heart of New York City at Times Square today to ask various people about their opinions on these new technologies that are coming out, whether that's self-driving cars or AI in general. Let's find out what people on the street are saying.

[00:16:16]>> In terms of AI, what are your thoughts on self-driving cars?

[00:16:20]>> I've never been in one, but sounds interesting. Maybe less crashes. Well, uh we have uh I have tried Whimo just really last month and in San Francisco and it's super reliable.

[00:16:34]>> Yeah.

[00:16:34]>> And the only difference between Whimo and a human driver, I think, is uh human drivers are more more flexible. We dare to sometimes, you know, disobey the rules, >> right?

[00:16:45]>> Uh WIMO is more reliable. So, right now, I think it's prioritizing safety than efficiency, >> right?

[00:16:53]>> It has to be safe. It's cannot afford to have any accidents at the beginning days.

[00:16:59]>> I think there is no there is no like a safest one because all depends who is driving or if that's like electric car and there only by by itself. Uh all depends about the company how how they made the the car or how their system works to to avoid uh accidents, you know.

[00:17:20]>> So there you have it. You've heard it from people on the street regarding their opinions on these topics and you can tell people are either very familiar with it and they have very thoughtful sort of thoughts about it or they're rather unfamiliar or they might have heard of the name and they're more curious about how it works and what it does. So with all that, let's go back to the studio with Muhammad Amodri.

[00:17:42]>> Welcome back to the studio. You know, there's lots of opinions. People have some sense of fear. Some say they'll get inside, others say they won't get inside. There's a lot of opinion on this. What What is this experience? What are the amenities? What are the comforts? Um I feel like the car hasn't been redesigned. Uh we're still sitting in the same way. How does this work?

[00:18:03]>> Well, it's a challenge for human ergonomic design where you want to make it as comfortable and psychologically appealing to a newbie who wants to be part of it. Um so think of it more like a train experience where you go, you can open up your laptop, you can start working, you don't have to worry about the driver driving the train. Eventually something like that would be required for a vehicle as well but you don't have to worry about that. The challenge of course is that um it's it's a lot of learning it's a a lot of innovation in the process and it will eventually and hopefully get there. Um you would also need connectivity between vehicles so that um when you have a mix of vehicles between humans and automated how do they communicate? How do they interact?

[00:18:41]That's a fascinating area. It's also an area of ethics because you want to make sure that everything is designed and planned ahead of time. How do you cater for uh you know god forbid accidents?

[00:18:52]How does a vehicle react to that? So those are factors which are still complicating factors in in the eventual adoption of these tech.

[00:18:59]>> Let's go to safety for a second. Um my understanding is that you know if someone is sick and needs to get to a hospital, someone is under the influence and someone's distracted by their cell phone, this should make things a lot safer.

[00:19:16]>> Uh in theory that should but what we have seen and what the data is telling us is that um a few years ago uh in the US there were about close to 40,000 fatalities 35,000 fatalities. I remember back in around 201617 that number went up and the main cult culprate it seems is the use of cell phones which means that even with more technology things are not happening as one would have hoped for. So the aim is that if you can somehow make this technology uh much safer um by making sure that even with a human operator you can have a level of control where the vehicle is operating at you know autonomy level four you don't have to worry about a human being because this interaction between a human and and the machine is what can create these cycles of confusion which obviously are not conducive for safe driving. And so either you have a fully human controlled uh vehicle with some augmentation from machine or you have a fully machine operated vehicle. So the the mix between the two can create those problems that that people >> there's a lot going on there. In terms of safety, I mean we've all seen or some of us may have seen some of these commercials where a police officer pulls over an autonomous car and who do they give a ticket to? Uh and and so on and so forth. when something goes wrong, I assume you can communicate with someone in these cars and they'll they'll pull over or do something and there's safeguards in place.

[00:20:39]>> Well, that's a legal question uh because it's a question that has to be resolved between the the state, the government, the the public and the insurance companies um and the automaker itself.

[00:20:50]So, that's a question I think which needs to be resolved through uh a more of a a legal recourse rather than through having you know an instant answer. The biggest challenge of course is that either a human is responsible or human is not and if a human is not responsible as a driver then the onus falls back on the the automaker in some some cases arguably. Um but someone can argue that it's not the automaker's fault because if you have a vehicle which is operating with supposedly human uh intervention then the human being is responsible. So there are certain legal nuances to that which have to be resolved. Yeah. that and then they're working on them and and as the innovation happens hopefully does get safer though. Now, Suz, so let's talk about robotics. It's your specialty.

[00:21:32]What breakthroughs need to happen in robotics for these cars to be truly autonomous? You know, we've been talking regarding these levels of autonomy and to not have a driver in the driver's seat. What needs to happen in robotics to get there?

[00:21:51]The hardest problem that we really need to solve uh with robotics is to be able to handle noise. The world is messy and the world is unpredictable. So for these systems to be able to work in real world, they need to be able to do what you mean by noise.

[00:22:10]>> So for example, um when a human is driving and a cyclist is by the road, we can easily get a sense whether the person is likely to cross or not. We have that sense that is a very hard problem for a robot to do. It takes a lot of training, a lot of machine learning time with a lot of data sets for it to be able to know when it is safe to drive by a cyclist versus when to stop. So some of these things are really hard problems that are really at the cutting edge of robotics and um autonomous vehicles and that is what uh the main focus is getting the right data training the systems on the right um data and making sure that we are putting in all the safety mechanisms for them to operate correctly.

[00:22:57]>> What happens with uh and how do we make advances here for for the robotics to get there? Where are we today? uh and and how do we make leaps forward?

[00:23:10]>> Um there are a couple of things. The first handicap that these systems have is that they need to be able to operate offline. Now if you are a human being, you are in an unusual situation. You are able to think and adapt easily. That is still a hard problem for perception systems. So being able to build systems that can adapt with very very exceptional situations that they've not been trained on. That's the first thing.

[00:23:38]The other thing is to make sure that they don't operate too robotically. That is the operation is smooth. It is not jittery and they don't make random sharp decisions and they integrate very very catchously with their environments.

[00:23:55]>> Do you think we're making progress and we're going to get there soon?

[00:23:59]It is a hard problem.

[00:24:00]>> It is a hard problem. Um but with the growth and scale of uh computational um resources, uh that problem is doable within a reasonable amount of time. It is still a hard problem. It is very hard to determine and give a number to when it will be solved.

[00:24:19]>> Thank thank you for joining us and thank you for sharing that perspective.

[00:24:24]>> Thank you.

[00:24:26]>> Well, thank thank you for your time. You know, this has been a fascinating discussion on robo taxis, taxis that don't have drivers in the driver's seat.

[00:24:34]They're on the streets today in some cities and more are coming. And you heard about the experience, the challenges, and the opportunities in making the world safer with these. But there's a lot of details still being worked out in this innovation technology. This brings us to the end of our open forum show. Our goal is to share data, share complex ideas, distill them down to simpler bite-sized pieces, and hopefully create dialogue in homes around the world where you can experience some of these things and keep the conversation going. Until next time.