 How many of you think you know autonomous car now? Could you define now? I mean, it's like when you take an exam, right? Professor can't give you a lecture. And then they ask you at the end of the lecture how much you have found that picture. And they roughly know maybe. Has any of you get a ride in an autonomous car before? No. So actually before I started working on an autonomous car I had no idea what it is. So I think at this point you are in a much better position than I was back then. So let's switch the topic a bit and I will present my personal journey of how I get involved in this kind of research and development and this technology. So before I talk about our company, Neutonomy is we are still a startup trying to tackle the most difficult challenge in self-driving cars that is urban driving. Why is it challenging? If you just drive on highway, it's usually very simple. You just keep staying in your lane following the front car. But when you drive in urban settings, there are people, there are j-working people, there are many cars that violate the rules. And you need to figure out how to handle these situations. So for our company we focus on the software part. We consider many pieces for the autonomous car. Study from perception means how do you try to get understanding of what the current goal is? Where are the other cars? Where are the other people? Where is our car in this world? And then from what you see, now you need to decide what to do. And that's what we call the planning control. So if you've got a lot of decision making, just like when you drive, there is a car stop in front of you. Are you overtaking or you want to stay behind? Once you make a decision, then you have to actually plan the motion. And then you need to actually actuate the car to follow that motion that you plan. And then there are many other things in the background as well like communication, fleet management, simulation, and so on. But the part I focus on personally is this second piece, which is the planning and also the control. That's where my expertise is. Sorry. And currently we are testing our software at many places, including the US, UK, and here in Singapore. And we tried on many types of cars, electric cars. Actually in Singapore, all the cars we have are electric. In the US, we try all the software on a hybrid car. And in the UK, it's also the IC platform. And we have no problem driving on the right-hand side, like in the US or left-hand side, like in Singapore. So that's about autonomy. Let me, from here, let me try to extend a bit how I get involved in this autonomous car business and this journey. Actually it started about 10 years ago. It's from 2006. That's when the DAPA announced the DAPA Urban Challenge. Has anyone here about this competition before? Oh, interesting. So the DAPA Urban Challenge is a race of autonomous vehicles. All the vehicles are completely autonomous. They give you a map of the environment, of the place your car has to drive. I think three days ahead of the race. Then five minutes before the race, they give you a thumb drive that says where your car has to get to. They give you the thumb drive, then they time it five minutes. You have, in these five minutes, you can do whatever with the car and then everybody has to get out of the car. There cannot be any communication at all with the car. The car cannot even signal what it's thinking, what it's trying to do. So the team has no idea what the car is doing. We only get that five minutes to set up everything, make sure that everything is working. We get out. The only thing, the only communication we have with the car is the remote e-stop. So this is with the DAPA staff. They have the remote e-stop that could stop the car when it's very dangerous. But usually before they stop, because if this is normal, we should let it go. And in that situation, we need to try to guess what the car is thinking, why it's doing that. It's safe to continue. Because if you run into an accident, then you might be pretty much done with the competition. So that's 2006. Actually, at that point, I was around the end of my first year as a grad student. I was pretty much having nothing much to do. So I was like, okay, I don't know what autonomous car is, but it sounds interesting. So I decided to, actually back then, it went like this. So there was an announcement about this competition. Then my advisor back then at Caltech said, oh, we want to involve in this competition. Now we need to organize some meetings, gathering students and people around. And he asked for some volunteer to, because we need to split up into groups. And each group needs some kind of coordinators. Oh, it's just like one hour meeting. I'll just do it. So I did it one hour. And then it became one hour a week. And by the time I know, it's like 20 plus hours a day. So that how it goes until 2007 is the actual race. So this is the car that we developed at Caltech. It's quite big. It's quite ugly. But back then, we think that this is quite cool. We have a bunch of sensors. This bumper is actually more expensive than the car itself because it has many, many sensors. Yeah. So that's 2007. We went to the race. And then after the race, I did still, then I was back to research as a PhD student. I got a lot of motivation from what we found at the race. Then my research was more on the VNV stuff, verification and validation. Then in 2010, Emilio started the Future Urban Mobility, the SMART that he mentioned. And then I joined because it's so interesting. I also like Singapore because it's so close to home. But my home is in Thailand, so it's very close. And then around 2012, I need to get back home, work in Thailand. And then around 2014, that's when Neutronomy launched. And I think at that point, I get to talk to Emilio again that we started this company and I was also interested in joining. So I worked part-time back then until I think around October of 2015. Actually, September, we bought the vehicle, our first vehicle in Singapore. And then in October, I joined. Then in November 2015, we have the first public demo of our software that happened in the US. And then just earlier this year, we finished converting our cars into autonomous cars. And then we went to LTA Safety Test. This is the test required for us to be able to drive around public road. So this is like when you drive as a human, you need to go through some kind of driver license test as well. So they have similar tests for autonomous cars. After March, then April, we get approval. So essentially we passed the test in March. And in April, we get formal approval that we can drive on a public road in One North. And then in May, we gave live demos to many journalists in One North. This is in a very uncontrolled environment because a bunch of cars can drive around as well. There are a lot of cars. And since then, we have been giving many, many demos. And each time we go out to a demo, we find many interesting situations because the environment changes every day. Sometimes it rains. Sometimes the sun is very bright. Sometimes the traffic is very heavy. Sometimes there are a lot of pedestrians. They're walking and so on. So the more we test, the more we learn about these situations we need to be able to handle. You can see the progress of the car. We started from this huge acrylic car into this kind of group which is much nicer. And now this car which is much cleaner and nicer. So that's like the progress we are making. We first show you... This is my first experience with the autonomous car. This is the DAPA Urban Challenge. The race was actually at the military base that is essentially abandoned. So there are still roads with all the signs, lane marking and everything. As I said, this is completely autonomous. Then there are also interesting situations for us with other human-driven vehicles. For example in this test run, there are like 10 to 12 human-driven cars driving around the loop and then we have to be able to merge into the traffic. There's also a situation where we need to handle traffic, the stop signs. There's also this big loop. We need to go around. There are many interesting scenarios here. We start from here, then we need to go through some narrow start chute. For us, because we have a huge car, we struggle a lot with this, trying to go through. Then there's also this narrow gate. Once you go down, you also find the narrow winding roads. There's a parking lot here which is almost full and we have to go into one spot. Then also there's this unstructured region with an opening in a fence that we have to be able to navigate. There's also a road with cars on each side that we have to be able to avoid. All this has to be completely autonomous. You can see how denser the traffic can be in one of these tests. So this is the video taken from a camera attached to our car. At this camp, there's absolutely nobody inside the car. So the car has to make the decision completely by itself. In certain situations, that bow asks us, do we want to emergency stop the car or not? So you can see that the road is also very narrow compared to our size. But in this test, the road is quite empty. But it's very narrow. Most of the scenarios are quite static. There are parked cars that we have to avoid and so on. I think this is where the emergency stop us because they thought we are heading into these fence. And then they asked us, do we want the car to continue? And back then we said, okay, we think it should be able to continue. And then they let it go. And then it's actually able to continue. So back then it's quite interesting and challenging. We have a team of about 50 people at a time. So we work quite closely together, spend a lot of time together as well. Let's see if there's something interesting. Actually the video is quite long because this road is very long. Let me show you another interesting one, which is the emerging one where that one, our car did very well. So for our car, dealing with static environment is very, very well. But when we need to merge into traffic with the narrow road, we didn't do it quite well. In this case we actually reversed in the middle of the intersection. This is because of many logics that we implemented. Essentially we don't want our car to get stuck at all. But then because you can see how narrow this road is and there's this carrier on the right. So that's how we try to get away from the carrier. But then on the left side there's also many cars. So we didn't do quite well in this case. And so from this, we actually learned a lot. It became the motivation of most of my work after that. So that's 2007. And then in 2010 we started the Future Urban Mobility. And was also part of it. So actually at SMART, I was working on many different things but all involving traffic systems. So autonomous car is one project. I was involved in. I also looked at traffic control. So that's more like the bigger picture of how you control the overall traffic. So over there we also developed like these are the golf carts. We convert them into autonomous car. And also run the public trials at the Chinese Garden as I mentioned. So here we ran it for six days, driven about 360 kilometers total with 500 visitors, 200 passports and many service. And I think this is some of the video from there. And have you heard about this and the chance to go? So there are kids riding. The nice thing about this is there are pedestrians walking around but there are no fast-moving car around us. And our car is still driving very slowly. So that's the part that makes it a bit more simpler than like driving around whatnot. So wasn't Emma quite happy with this? Have a very good experience? I cannot hear the audio here but this lady is saying I was eating the car and the car was staying away from the tree. Yes. But that's kind of the point. Yeah. Well, it's a theory that doesn't happen. When somebody crosses it, it would stop, right? Yeah. And this lady is actually saying that this would be very useful for her. She doesn't want to walk too far in the park. So having such a service available in the park would be very beneficial, especially for the senior citizens. No. Sorry. This is because I have to turn this on. Okay. So that is smart. So finally, we posted a team from smart moved over to the company once we know that the technology is a bit more not that we're sure, but it's still to the point that we can try out in a more larger scale. So this is the part of the public demo we give around why not. Not sure if it's the same video. That day we gave so many demos. I think we do like 20 plus runs. And so this is one of the videos from there. You can see that this is part card that we have to avoid with some pedestrian walking around this card behind us. Yeah. It's quite very interesting riding around there. Until then, we learned something by just driving around. There was one day, there was a car going this direction, which is completely wrong. So we saw pedestrian, we stopped for them. In this video, you can see that the car is actually driving around. Manually taking over around. When we turn here, there's a parked truck. It was quite I mean, we are still developing the technology. We were running the latest ever court that we are actually in the beginning, we were debating whether we should run the version we know it was running, but it's less exciting. Or we should turn on all these new features about avoidance and everything. And then we said, okay, let's just make it a bit more interesting and just run the latest software. Turns out that it was doing quite nice with all the avoidance following Yeah, but some of the complaint we get is that the ride is a bit jerky. So with that comment then we try to improve the performance of the car, making the ride smoother. So that's one small loop. Actually in this small loop, we usually find many interesting behaviors. So this happened around May. And then we also, we actually keep testing. As I said, we'll find more and more interesting situation. This is another term we run different days. These days, there is a series of car parking and we need to be able to avoid. So some days the area is just crazy. There are so many illegal parkings. Many