 Okay, so good morning everybody. My name is Bernard. Before I start, I will just tell you a little bit about my background. I'm a trained theoretical physicist from Cambridge and my PhD thesis was about finding extra dimensions through gravitational waves. So you could imagine my excitement when gravitational waves was discovered. But one part of my life was actually spent here in the Singapore Science Centre. When I was between high school to junior college, I actually spent a very significant amount of time interning here every summer and winter. Or maybe June, December holidays. And the reason for that was because in the 1980s, the Singapore Science Centre is actually the only place that has one of the biggest, most important classic called Gravitation, written by Wheeler. And it is one of the most important books in general relativity. And that was how it got me into becoming a scientist. And just to add one more point, subsequently I went and worked on the Human Genome Project. And the reason why I could work on the Human Genome Project was because of one important biology lecture that was given to me by a lecturer in National University of Singapore. And his name turns out to be Dr Lim Teck Main, who is the director of the Singapore Science Centre today. And it was actually that lecturer who was actually a Socratic lecturer. He never actually tells anything. He actually tried to get us through the whole process of Socratic method, actually got us to learn. And actually through that whole experience, it got me into learning things in physics and biology, in economics and management and in technology. So what I'm going to tell you today is a little story about how Singapore Post, which is actually a commercial public-listed company, worked out a drone delivery system within three months. So the story is actually a little bit interesting because it started on the very first day of my work. So like any other people, I started work on the 2nd of December 2013. And I of course put out a tweet saying, I'm joining St. Post as the head of digital. And one of my friends, Smini, decided to be cheeky because it was actually the same day that Amazon actually launched the Amazon Air. And if you do not know, Amazon Air is actually the drone that was started by the Amazon. And you see, St. Post Air I assume is the first project. That was very cheeky because of course not, I wasn't hired to do that. But this tweet I deliberately kept it because I know at some point some things are meant to be and eventually it will come to this project. So what happened in the next 1.8 years? So the next 1.8 years I was trying to solve very interesting problems within St. Post with digital products and services. One of them is called the PopStation. I don't know how many of you know what's a PopStation. It's actually a parcel locker. It's one of the services that every day if I look at our social media dashboard, it's the ones that everybody is happy with the service. In fact, our dream is to have PopStations everywhere on that. And we actually, it's one of the only few projects in St. Post where we actually open our APIs to e-commerce people so that they can actually use that and actually open and close a lot. But one of the interesting things that come out about it and this will come to the later part of the story was that it was actually the first implementation of Internet of Things. It's that you can actually use an iOS or Android device. You can actually air unlock that particular parcel locker with your phone. And that itself actually at that point in time we were actually worried whether the technology was actually ready for that. In fact, I've the challenge a couple of claims in security and we're actually using some of these open source software to actually get it done. So that was one of the things that we learned and of course the Prime Minister actually tested this product in one of his walkabouts. And then after that, we also redesigned the Sand Machine and I repurposed the mobile app which was 150, when I was there it was only 50,000 downloads and then by the time I redesigned everything, it would make it really cool. It's now close to a million downloads now. Of course we'd better find star ratings from a one-star rating. And of course the most difficult project as the head of digital was actually redesigning the Singapore Post website. Yeah, it may sound as though all the other problems are simple but it turns out that now, but solving these problems actually gets you towards an innovation challenge and let me tell you what happened. So while I was solving all these technical problems, there is also the Urban Logistics Challenge. One of the biggest things is that we're going to have a big e-commerce uprising and increasingly we're getting traffic congestion, right? I mean a lot of you these days if you're driving on the road you will see a lot of highways are being jammed because the amount of infrastructure we have just can't resolve. So we actually need other things to solve that and of course we also have manpower concerns. It's not easy to hire a courier delivery man or a driver these days. So all these have sort of amalgated and sort of came to us to think a little bit about autonomous vehicles, not just air but also land. So at a point in time I was actually working on an internal project to think a little bit about drones and self-driving cars and what happened was my boss, Wolfgang and Mr. Wu, the postmaster of SingPost, had lunch with the CEO of IDA and she was like suggesting, hey, why can't you guys just fly a drone? So the IDA is actually a regulatory authority. So they came back, they told me, Bernard, go and find a project to do for this. And so I decided that, okay, let's talk to them. We have some other projects, maybe less ambitious than a drone delivery project, like having a smart post, maybe a smart post box, maybe trying to do big data, but no, no, no, have to be a drone delivery. So we were trying to think about this in a few ways. So one of the things, Objectives, was actually to support the Smart Nation initiative. But the second one is actually, this was actually the time where we are actually about to leapfrog ourselves from just a follower to an innovator. And the next part of it, which actually came from both the CEO and myself is that we were thinking that if we were to do this, it's actually not going to lead to any commercial applications so soon. But what we would do is that actually we would open the whole regulatory framework for drone delivery. It would help the startups, it would also help the corporations in the future in order to have that to happen. And then it basically helped us to expand the ecosystem, not just us, you know, for drone delivery to actually make it happen. So I decided to, I went to the, I went to talk to different drone delivery companies and decided to work with the IDA design labs, which is actually a government lab inside IDA that actually do a lot of cool technologies. And you actually work together and basically come up with the things. But knowing how government works, I think a lot of you know, they move very slow. So it's a big problem, right? So what I decided to do was, a little bit tongue-in-cheek, tongue-in-cheek in the sense that I told them, so it's now June, there is a World Post Day in October. So just to give you a sense about postal services, every year there's a World Post Day on the 9th of October. We celebrate the postal service, we celebrate postman, we celebrate front-liner, post office stuff. So St. Paul's used this as a tradition to actually honour our low-income workers and et cetera, and we actually do a lot of service thanks on the day. So this year, I decided that, hey, why don't we tell the world we did drone delivery? So I forced the timeline onto my government counterparts, I need three months to get this drone delivery system done, and this is where we're going to go. But I don't want to just ape everyone. I don't want to follow what Amazon did. I don't want to be like DHL, fly something from point A to point B and say, hey, we are doing drone delivery. I also don't want to follow what all the US posts, Swiss posts, Fin posts, they are all basically following the same recipe. Just fly something from point A to point B, which is totally defeating the whole purpose. If you were to truly innovate, you need to truly think about what does drone delivery actually mean in the postal world? Basically, traditional postal delivery is very simple. Point A to point B, a sender, a recipient. At the recipient side, you've got a key, you have a mailbox, only you are the intended recipient, unless of course there's errors of the delivery. But in a drone world, you can't just fly from point A to point B and drop a payload, like the way how DHL does it. And then you expect that the person who's collecting that is the rightful person that actually collects that. And that is where we decided that, hey, if we were to do this drone delivery, why don't we show exactly how this would work in real life and how could it be operationalised in the future? And we have to take a real problem for ourselves and say, we are solving an actual problem that Singapore's really have for drone delivery. We are not going to go out and, I mean, it's very easy to go to a shop and buy a drone and fly from point A to point B and say, hey, we did drone delivery, right? So it turns out that in... I will tell you the problem later, but let me just go to the next slide. How far actually am I... I mean, the answer to the question is, how soon will I be able to deliver pizza to your place with the drones? The answer is not so soon. There are a couple of things. First of all is the unit economics. The cost of commercial drones today is at sub-1,000, but based on Moore's law with drones, to the point where it reaches sub-100 dollars, then there is an issue. Then we actually have a possibility of having commercial consumer drone delivery. The second part of it is actually regulatory costs. Just to let you know, Singapore legislation is that you have to pay $70 per flight. Come on, every week I send only 50 letters to pull out will be $0.30 per letter. So the unit economics just simply doesn't work out at this point. There is the manpower challenge, right? We need to actually train the workforce. And even if we have to have everything autonomous, whether it's self-driving, whether it's drones, we still need to have manual overrides. And then the last bit is regulatory challenges. Now, here's a little interesting problem. The technology is there. It's all there. The problem is actually with the regulators. They are not actually being educated with that. Now, let me give you a very simple problem. I was actually having a lot of fights with the Aviation Authority to get the drone fly. And one of the most interesting things is when you start having an Air Force captain who is a regulator with training in engineering, telling a physicist with a PhD how a drone actually flies like an airplane and not like a helicopter, you know there's a problem, right? And he keeps overriding and trying to tell you, look, you're just a physicist. Oh, yeah, I'm sure. I am a physicist, right? So there's a lot of things that you actually need to educate. But I think through that process, you start to understand how the regulators think about security, things about a lot of very silly nuances, and then you have to challenge all the way. And just to be... even though we work with IDA, we almost couldn't fly. So after doing the first... we actually came up with the prototype. We configured a couple of things. I'll talk a little bit about the specs that we did. And then we actually did some trials. And then we finally get to the September. Now, something happened in September. Can anybody remember what happened during that period of time? Haze, right? So I also had another fight with the authorities saying that you can't fly an unmanned vehicle in a small area. Sure, unmanned, I know where it's flying, right? Yeah, so in the end, I compromised and said, look, okay, if it's 100 psi, we will fly. Now, it turns out that I have to fly sometime in end September because one of the technical leads that we were working with IDA is actually leaving. So the problem was that we almost couldn't fly and the whole project would go down to smoking. Now, when we went to look at the drone, we decided to do a couple of things. So there's no point. We looked at all the closed-source solutions out there, like the DJI, they have no open APIs. There's nothing I can hack into it, okay? So we decided to look for a particular open-source platform and we decided to use the Pixhawk Steady Drone platform. Now, one of the things that we actually did that is slightly different is that remember the story I told you about parcel lockers and opening the parcel locker with a phone, right? So it turns out that there's actually an authentication procedure in the drone delivery flight from point A to point B. At point B, if I'm standing here and I have a mobile phone, I can actually use my phone to authenticate myself that I'm the intended recipient of that particular package. So when the drone flies to point B, it will actually hovers around and all I need to do is to say that I am the intended postman to receive this package and I press a particular authentication and then the drone will actually land and we could pick up that package and then I know that this is the intended piece. So at that point I need an open-source software and I just don't need the open-source software from the drone side, but I also need something on the phone side. So we decided to try to use an Android operating system. We ran it on Xiaomi Mi 4 phone. Actually, we could choose any other Android phones. It's just that the developer who was working with us actually loves the Xiaomi phone. So we did a couple of changes to that. The wings is actually usually I think all the commercial drone is usually using plastic wings. We actually turn it into wood wings because of certain fear of airwardiness. We actually 3D printed the entire mailbox and the payload is actually 3.8 kg. There's actually including the 500 grams of payload and the parachute and mailbox and it runs about 8 meters per second and it's approximately about 30 kilometers per hour, but a parachute to actually when we fly across the island and it drops into the sea okay that's not the problem but when you fly across the land and there's a prem nearby then it's better to have a parachute because this is like a flying blender. It could actually hit somebody. So we put in a parachute as well. We also put in a little bit of a GPS so that we can actually track where the drone is and we actually had to get a special frequency from IDA to allow us to have control to actually communicate with the drone because it's actually going beyond the 1 kilometer distance because most commercial drones today only could last a kilometer as such and we're actually going beyond 2 kilometers. But that particular special frequency will change with once our mobile handsets could go into 4G and 5G mobile carriers. Now, I want to show you some of the things that we did with the test flight. So if we didn't just take something, hey, build something and fly, we actually have to convince the authority. So we got all these Air Force captains and showed all our 12 flights with the endurance flights. We actually test flight until the battery runs date. We did the GPS guidance and then we actually run it across from the airport. So this is one of the two places, one of them is Labrador Park but I'll show you another interesting part of the delivery, the trials that you actually see. So this was actually more for our own internal to sort of guide us where we are in the process itself. Now, this is the fun part. Okay. This is one of our last test flight. We were actually facing to Santa Rosa Cove if you look at the edge on the other side. So there was one fine day, we took a week day and nobody knew. We flew something because we were a bit worried if somebody would take a picture of this drone. So we have to package it to look a little bit different and then we try to fly it with a payload. But one of the interesting things of doing in front of this bay area is that you're actually getting an airwardness that in view of a very high wind speed, this drone can actually withstand the wind on that and it's actually flying around it, you see? It gets around and around and well, here's the fun part. We were trying to be cheeky. We bought a DJI Phantom 3 drone and we got them to chase and the moment the DJI drone and this drone fly, the DJI drone couldn't chase at all. In fact, the DJI is like a tortoise and this guy is like in the analogy of that. And we have to slow the drone down so that the DJI can actually capture. You can notice the blades there. You have to reduce the speed significantly and then it gets back to where we were on that. So we decided to choose this drone fly to pull out Ubin. So a lot of people must be asking why. It turns out that we do have a real business case. We have a postman that goes to pull out Ubin three times a week. He's already in his seventies, okay? And we have actually recalled him twice from retirement to help us to deliver letters to the island because this is the and we couldn't find a successor to actually deal with that problem. So the way we were thought about this process is that, okay, if we were to make drone delivery and reality of delivering letters and packages to pull out Ubin, then what we need is that what we want to do is to actually get one of the residents within Ubin to be actually our postman. So we actually land at a particular location. The postman actually collects the resident within Ubin and actually distributes within the whole island itself. And so we can actually time this because there's actually a delivery base near that area there. So we chose different paths. Now here lies the real difficult and a lot of you might not know this. It turns out that the path that I took is an impossible path to ever get approved by the authorities. And let me explain why. First of all this path looks very nice clear line of sight. But it turns out it's in the heart of tree aerodomes that tree airports that actually uses this particular path. The first one is Changi Paya Lebar which is a military airport. Of course you know anything to do with military, oops Bernard you're not allowed to do that. And then there's a little airport private airport. So we fought a lot of things on there. We fought things like well the drone is not flying above a certain 70 meters so we should never get touched with a commercial airplane. And then you start having these Air Force captains because they're really worried that the drone will fly into one of the aeroplanes from each side and each side. So okay the military airport is closed on a weekend. Great. So let's, so we should be allowed to fly on those days which the military is not flying right. Then then the next excuse they come up is what if it veered towards this side and it hits an airplane on Changi airport which is completely possible or it goes into this side which is actually much more interesting. Your drone flies across border into Malaysia territories. In in simple rule, this is basically means an act of war right. Yeah. So you can see how difficult it is to get it approved right. So in the end I compromised okay I would tell you that until the day before I fly I got the approval on a Saturday 2 p.m. and just to explain to all my gig friends why I didn't show up in last year's gig camp. That was the day I was waiting to get the approval for flight. I actually almost wanted to come now and get help from Harish to talk to Vivian Balakrishnan to get the flight through but I decided that if I were to do that it will actually fail the spirit of the law. The spirit of it is that we want to work with the regulatory authorities and we go by the proper channels to actually get the approval done. Okay. So we challenge upwards. We are challenged to the next higher authority above the aviation authority. I said look this guy is just being unreasonable. Here is all the email correspondence here the data that I've shown and please be my referee and they were very fair. They were being very fair. I compromised by basically reducing the battery of the drone to 80% so that when it veers off actually we had a Joe Fencing strategy that if the thing starts veering off within 0.5 meter radius or from the path we actually would shut it down. So we were actually very, very in Singaporean language kiasun about this. Literally we will shut down the drone once it veers off in a certain path but in order to actually make them feel better we just said we will reduce it down to 80% of the battery so that even if it veers off and go this way and that way there will be a drone flying across the streets of Johor and it won't hit a commercial airplane. So that was the compromise that we actually gave and thanks to the help of the Ministry of Transport who actually came and actually be the referee and show us what is the deal process with the authorities we got to fly. It was actually on Saturday at 2pm actually I wrote two emails to my boss. One is I failed because blah blah blah blah. The other one is we are going to fly. So we got to fly. So we were also lucky. The next day we decided to do it very early in the morning which is I think 9am because the PSI was about to go 200. So the moment we finished the flights, we actually did two flights, we actually managed to get the 200 PSI start coming in and then for the next three we resolved 200 PSI. So we were very, very lucky. So this is the whole team that did this. It's actually not just me. There's a lot of people that's involved. There's the IDA design team who I was really glad. I didn't realize that in Singapore Government there's a group of real good technology open source gigs that I can actually work with and this is one of the best projects that I've ever done. Now I want to show you something but that's just one more thing. So this you won't be able to see because we didn't show this and BBC Reuters did ask me and I didn't give them the footage. So this is the actual raw footage of this that we needed to keep to have that. But I'll show you to the part where we are about to take off first. So this was the second flight that we did. The first flight we actually did it was successful. It go back and then we did the authentication and LAN. One of the things we wanted to do is actually we mounted a GoPro camera but the problem is with all technology every time you do it the first time you forgot one button and it didn't take the footage. So shucks we need to get a second footage and there's only one more hour before the 200 PSI is coming in. So we tried and we actually did and by the way the second time the guy made the same mistake and pressed the wrong button and the footage was lost. At first we thought it was lost but then luckily because they've done this so many times there's a particular software that they have which is open source 2 that allows them to recover the footage and so we actually got this footage for today. Now one of the things I wanted to show you which I actually never appreciated about Singapore going to the going to the offshore islands is I don't know whether you see this there. These are all called Kailongs. It's actually things that are built on wooden stilts. It's one of the old heritage of Singapore where people still live in the seas etc. This is one of the only few views that you can actually see that on that and we're actually flying right in. We actually located a particular location in Pulao and it was quite funny because if you look at just the airspeed, we're actually controlling the tilt of it. Sometimes we're trying to balance it on that. So as we go into the Pulao will be inside and then we will at some point it will hover and then you'll feel that it's actually hovering and then we actually try to stop the thing and then we authenticate, we actually fly. They actually authenticate the the person below and doesn't it will fly back but we actually did the authentication and then it starts to slowly hovering down as well. Originally, we only actually need one person to manually override it but because in order to get my regulatory approval again, I needed three people to do that. So this is one of those footage that I thought would be very interesting to share in the spirit of open source. Now in the spirit of open source, we took all the data that we learned from this exercise and we give it to the government for free on a couple of reasons. One reason is because we need them to understand how difficult it is to get a regulatory permit from a drone across two locations. Second, I think both my boss and I feel that if we were to share this with them we would be able to get them to make better legislation for us. As a private sector if we are going to have a better ecosystem for startups and companies, this is the way to go. So what happened? What happened was we were lucky because of the way we think about this drone delivery, we got more coverage than all the other posts add up together. We even got a very nice line from Bloomberg Singapore post like Amazon test package delivery by drone. But I thought one of the interesting stories that was told to me recently was apparently and actually this is actually a public Airbus and CWAS signed a MOU to do drone delivery. The reason why it happened was because of our drone delivery. Airbus saw what we did and said hey, actually we can work with these guys, bunch of government guys to actually get this thing done. Good luck by proving airwardiness. But I think they will have a better shot at we do improving airwardiness. But a story must have an ending. So, you saw the tweet that was given to me two years ago, right? I actually saved the tweet, etc. So I wrote back a tweet storm to my friend two years later and in order, because I work for a public company, so I had to say I need to get clearance from COPCOMS blah blah blah, but here's the solution two years later and I'm going to end here. Thank you. Alright, thank you very much.