About two billion people that are going to move into cities by 2050 and with that growth, the demand for efficient transportation is going to increase dramatically. In an era where we’re already seeing inefficiencies in urban mobility having a massive impact on the economy, public health and environmental health, it’s hard to imagine a future of transportation that doesn’t border utter chaos. Cognizant of these projected problems, Assaf Biderman, is working on solutions that harness the power of artificial intelligence, robotics and other technologies that are already within our reach. Assaf is founder and CEO of Superpedestrian, founder of the Senseable City Lab at MIT and an awesome guy to learn from. I’ll admit, I have been dubious about the rentable scooter business, but Assaf has me convinced there’s an important place for these things in our cities. If you have any interest in urban mobility, this conversation is important.
You’re still at MIT, but you don’t have to go anymore because no one goes to work anymore.
The whole lab has been removed since March. I’m still on the board of the lab. I spend most of my time at Superpedestrian.
Is the Senseable City Lab still going?
What are you guys trying to do?
Senseable started in 2003, 2004, where the goal was to say, computers are becoming part of everything. They can emit data. They can act on data. You can embed them in your environment. That allows us to completely change the way we study design and impact cities. Some people call this field of smart cities and I don’t like that.
It is because there’s no such thing
People are smart enough, but there is a lot that you can do. You can discover new things about how people organize themselves and about how it flows through the city, energy flows, waste, the things we consume, people, and communication. A lot of that can impact how you design them and how you manage them in real-time. It’s got a lot of value. It’s one of the largest lab fields. I’ve been doing work since 2004 in partnership with cities all over the world. Those are big city partners and a lot smaller. It was funded by corporate for the most part and more will survive by long-term brands, but most of the money came from corporate where cities volunteered themselves as a subject matter and tell us about what problems they care about. Probably they want to look at together with us. We use the bigger money from corporate, all thrown together into one pot. We basically manage the deployment of dollars into research areas that we care about and the cities care about and the consortium that the management cares about. Most of the time, technology surrounds machine learning, robotics, various types of analytics.
For example, when you think of the seventeen-year history or something, what are the things that stand out to you as examples of what that lab is doing so that I could understand?
The impact areas that we care about are the stuff that makes cities function better or worse. We look at a lot of transportation, and probably half of those are transportation, whether it is dispatch algorithms to global taxis that we’ve been working on for many years. There’s quite a bit of knowledge there that’s generated this whole micro vehicle angle, which is what Superpedestrian is spun into.
How do you define micro vehicles?
These are tiny vehicles that take vertical space. The key thing is you got to take much less space on the road than a car does, but the longer answer it depends on the occupants. We want to make sure that we are able to get a lot more people on the road. There are about 2 billion people that are going to move into cities by 2050. There is no way that these people are moving. Cities are already overbooked so 1.16 people in a car, which is what we do today, like Sedan don’t cut it. Think of something else where the utilization is a lot higher, either a tiny vehicle for “1.16 people” or some way to decline transit with these other modes. It’s classically called multimodality. Cities have done this for over a century with subways. You walked into the subway. Now we want to extend the reach of these systems or any new modes on either end.
Two billion people are going to move into cities by 2050. There is no way those people are moving if we don’t revolutionize transportation.Tweet
I have a zillion questions about this.
That’s one thing about transportation. We’ve got a lot of work on housing. I work on energy, waste production, and communication. How do communities lay themselves out in space, which touches on the essence of why we come together probably in cities, in the first place? It is to get access the diversity to human capital. There’s increasingly work on the intersection between biology and cities because you can learn so much about how we are impacted by a built environment or how it should be better designed by studying, for example, the collective microbiome of communities. It’s public health that provides that like the spread of pandemics. It is quite a bit of work on that in the last several years with robots that eat your shit and tell you something about the collective gut. It’s been spun off. They started doing well when COVID came around.
I’ve heard multiple pitches for companies that want to sample sewer systems to figure out where in the city there’s COVID or whatever pandemic.
There is this company called Biobot, which we spun of Senseable from the project we called Underworlds. I’m looking at it for a while at Eric Alm’s lab. Well trying to amply signals in the sewer that relate to that.
You said you’re on the board, but are you actively doing anything at the lab or you’re pretty much busy with the startup?
I am, but much less than intensive. A startup is something you want to put yourself into completely.
Coming at it from a different perspective, at Intellectual Ventures Lab where I was working., we created a group called Institute for Disease Modeling. Essentially, what the team was doing was computational modeling on the spread of disease. Mostly third world stuff, but what happened was we created computational models on things primarily like malaria and big infectious diseases there. When the first Ebola outbreak happened, 12,000 lives were lost. When the second Ebola outbreak happened, twelve lives were lost. In between what happened among other things was we were able to do computational models to optimize the ring vaccination campaigns so that we could quickly contain those outbreaks. That’s a two order of magnitude improvement on lives lost. With infectious disease, it could be a pandemic candidate.
A lot of the gains came because we were able to use computers to help us make better decisions. I use that in my mind as an example of how computers with these modern tools, computational models, big data, machine learning, network models, being able to help humans make better decisions about what to do with large complicated problems. Things that our intuition would fail us. That’s the rough thesis I have in my head about what’s possible. What I imagine is that you have a lot more experience than me with the Senseable Lab going and figuring out for cities. Exactly that because in my mind, in the future, even the near future, we should be able to use these types of tools to help us make better decisions about city planning, transportation, infrastructure like sewer systems, water, power and all those things.
We should be able to make use models to create thousands of possible futures and let us choose which one is the best from that. It seems like that’s the work you’ve been doing for a long time, in some sense. What I want to do is figure out if you have examples from that work to help me substantiate this view, help people see the potential and how we make decisions for situations, where asking an expert isn’t going to get you the best answer because we’ve tapped our tapped out on our potential with that type of decision-making process.
There’s plenty of examples. When we started this work, it was a few groups around the world looking at this and there was very little data to go by. We had to do a lot of data, to partner with companies that have data running through them for other purposes, cell phone companies. We worked with a lot of vehicle services and also the computational capabilities, the analysis, the science behind the analysis, the mathematics was it a difference. Some of this is being put to practice. It’s in a very different state. I still think we’re fairly early on. I’m not sure that the biggest value to be had from this cyber-physical city is in the ability to somehow centrally make decisions that are smarter. When it comes to pandemic, you definitely want that. That’s the great rule, catch to spread, know how to cut the chain of contagion, what we’ve all learned to think of a split flattening the curve.
All this stuff is perfect for centralized decision-making systems where theaters can be effective. However, if you think about a lot of the things that are happening in the city. People, the citizens, the actors themselves are the ultimate decision-makers, how something ends up behaving, how you navigate yourself through the city by what you throw away, where you work, and where you live. Almost all the decisions about the way a place functions played by the people who use the place, not the people who mattered. I think there is an important distinction that I want to put out there first. It becomes interesting when part of these computer models or computer analysis becomes open and information brings a citizen as an ultimate decision-maker, so that citizenship could become smarter.
It’s a good, amazing dream.
There’s little stuff that we’re already doing, but there’s a lot more that we can do.
When you say that, what I imagined is like, Mumbai versus Seattle or where in Mumbai, citizens are making all the decisions. There’s nothing central going on. That’s affecting the city. As far as I can tell, Seattle, the city is making decisions, largely poor ones about everything. The citizens aren’t making a lot of the decisions I don’t feel. Am I wrong about that? Do you see a spectrum there?
Think of the things that you do from the moment you get up in the morning until you go to sleep, how much of that is being dictated by somebody else? There are very few. It’s in your immediate environment that predicts how you move, where you eat, what’d you buy, how you consume, pollute, contribute what you work on, and all that stuff. Transportation, housing, and education are big pillars of cities. They are navigated by cities. At the end of the day, people are driving these pillars. Let’s leave it at this abstract level for a second and try to focus on it. Let’s talk specifically about transportation. Now, we make choices about transportation that are driven by where we were, where we live, and where we’ll spend our time. We might choose where to live based on the availability of transportation, how bad is traffic. Is there a rail going to where I’m going to move to? At the end of the day, there is the impact of the decisions made by the Central Transportation Authority. It can be a mess. It is something we’re planning for an alternative can be a mess. Let’s think what’s positive. We have a pretty good understanding of how demand has been shaping for transportation. It has been growing and growing. It hasn’t had a flat year since 1984.
Except for this one.
Cities have been completely overbooked on the streets, but demand keeps piling on, so what are we going to do? Many cities have been focusing on mass transit. For the past 100 years, we’ve seen subway systems pure around the world. They’re expensive, but they’re very effective. Many cities do not have the ability to put a subway system yeah together. Those who have them, can’t expand them as effectively. We’re seeing examples all throughout the world, but almost every developed city very slow expands the subways. Not only this, when you expand it, the city develops around the subway system in a way that it immediately creates new demand for transportation further in the peripheral. The subway system reaches a certain neighborhood then that neighborhood will be more integrated with the city.
Farther into the periphery, you’ll have new developments where people will buy cars and end up driving into the city from there. Historically, how do you plan that? You ask a bunch of households where they go to and from and create what’s called origin-destination makers. Now, we can derive those matrices with much higher procedures, whether it is through cell phone data. There are quite a few other ways. Think of the data that flows through an Uber-like service, a lift-like service, a transit authority, a taxi service, a micro-mobility company, and a cell phone company. When you put those together, you get a pretty good understanding of what’s going on in the city from the demand perspective. The insurance companies are increasing.
You’re looking at that so you get a pretty good understanding of demand, but at the end of the day, you can’t change the supplier of transportation real-time. The cities are made of concrete. It takes a long time to change those big things like infrastructure. There is planning that happens as a result and the planning is quite long-term of the transit system. You do these demand surveys and the city ends up making decisions, 10 sometimes 20 years into the future in terms of massive investments or even management. Now, the question is we did an explosion in urbanization and the complete over-consumption of transportation, especially in the urban part, but also in four-doors in the Bay Area. It’s congested throughout almost all week days. The same is true for our latest industry in London, Beijing, Jakarta and Bangkok. What do we do now?
If historically, we learn how to plan transportations that have walked to the subway and then you’d walk again, we’ve learned how to do this. They are all planners who specialize in this. They work in City Halls and they can be using the latest and greatest data tools or the sources, but still, you walk to the subway. There were some ways systems evolved very slowly. We added the Bus Rapid Transit system. If you want to expand the regional systems, there need to be new modes of transit that can let you go farther than what you get than by walking. Think of the KickScooter that we’re seeing a lot more around, shared e-scooters, shared e-bikes, and mopeds. One person covered vehicles like a tiny car electric. All these are going to become a part of a multimodal, very flexible transit system. That’s most likely going to be part of mobility or urban mobility systems into the future.
Do you think these scooters that we see for rent on the street are here to stay in cities? Are they improving things?
In some ways, yes, but let’s take a step backward. First of all, I think from a form factor perspective, what matters is that people like them. They find them useful. They have improved. They will improve a lot. That is what Superpedestrian is about, that technologies that make them safer and more manageable, but you’ll see multiple forms. This one form factor addresses a certain type of user, which is those short 1 to 3-kilometer trips.
They’re not great for groceries or whatever.
You can’t hold stuff on them very easily. You don’t want to go too far, but they’re great if you’re wearing a skirt, if you’re wearing a suit, or if you want something non-committal and they’re fun. Most importantly, people seem to like it. In 2019, there are 250 million trips made on those and it is growing. That means that there’s something that people are willing to do instead of using other modes. The question is, what does it come at the expense of? Is it replacing walking? Is it replacing subway trips or other transit trips? Is it replacing a ride-hailing trips or private carts? There are more and more research that shows that at least a third of the hour on average is already thinking mode shift away from private car use or ride hailing what cities crave. Elected officials are being elected into office to do something about traffic problems, which wasn’t the case, get cars off the street.
If you have one mode, the scooters, which takes care of the short end of the trips, we’ll see something else which takes care of the medium-range trips, electric bikes, for example, or even longer the moped. Perhaps a slightly different type of trip would be something where you have covered vehicles and you can carry your shopping or your child, all of this under the mandate of being small. It takes little space on the road and it is electric so that it doesn’t pollute. The question is, can these things combined so that you now have an option for coming from the periphery of a city into a transit hub or the city center? By combining walking with a scooter with a car or whatever, any kind of combination thereof, that’s multimodality. We know how to plan this. Now, we have the data system that can allow us to merge these new vehicles. From the academic perspective, the planning is pretty well figured out. The vehicles themselves have been lagging behind the engineering that the technology of businesspeople making them a possibility has been lagging behind planning of this.
I’m glad to hear you say that because first of all, you said a quarter billion trips, but that’s not much.
Don’t quote me on the number, but compared to something on the order of 20 to 25 trillion passenger kilometers, driven by individuals in cities every year.
We have a ways to go. That’s a way of measuring ran from a mobility passenger, a trillion passenger-kilometers per cab.
If I am not wrong, there are over 22 trillion passenger-kilometers around the world and the way you want to measure it is by passenger-kilometers per cab by a trillion passengers.
You divide 22 trillion by global population.
We’re expecting 60 trillion to 70 trillion passengers-kilometers by the middle of the century. It’s mind-blowing. The question is, how do you provide that in giant vehicle space? There’s not going to be room to support that. If you don’t supply this, it is not just about our life is going to be terrible or understanding traffic. It’s other things. Eighty percent of the world’s GDP is provided by the cities. If you don’t keep supplying it with transportation, if you don’t keep up with its growth, you get stuck to the GDP. Environmentally, there is an impact. In terms of anything from productivity to public health, the impact is major.
Did you say 60 to 70 trillion on what timeline?
By the middle of the century, by 2050. These are predictions by Arthur D. Little and the Bureau of Transportation Statistics. There is another similar magnitude. The vehicle’s been lagging behind It’s an engineering problem and it’s a business model.
The business model problem is clear. My experience with multimodal is that the transitions are expensive, especially waiting for a bus, waiting for a subway, getting off the subway, looking around for a Lime scooter, or whatever. The transitions are expensive and it makes me not want to rely on it. If I can afford something else, I’ll choose it and I can usually so I pick a car.
Dependability is key. You need to be able to make a choice of transportation mode or modes. If you don’t know that it’s going to be there for you in the morning, you got to be late for work. Once or twice, and you’re done. That’s key. That’s a lot of what we ended up focusing on Superpedestrian tend to do with the reliability of service and dependability. At the moment, it is still in juvenile stages, if you look at scooter services out there because the vehicles are wrong vehicles. The technology within them is not the right one.
What’s an example and how it could be better?
They failed to offer. They come there and it’s broken.
It’s driving me nuts. I get so mad trying to fire up a Bird scooter in Santa Monica and it’s just pathetic. The battery’s dead. I didn’t park it in the right place. I got to drag it to some place where it’s geo-fenced properly.
The fact that the technologies behind them are not fit for the purposes, making it so that it’s expensive for the operators to offer a service that they can’t offer enough, in any way, that’s economical. If we want to see success in cities, we need to see sometimes two orders of magnitude, more of these vehicles, which means they need to be a property. They can’t be just thrown everywhere. Technology is required there. How do you do a much better location? How can the vehicles autonomous needs enforce their own geofences in a much more precise way, which is something we have investing on?
Manage the vehicles, but then economically, if your vehicles fail all the time and your cost of offering a service is driven by the amount of people you have to throw at the problem or keeping them up and running, replacing parts in them all the time, which is what scooter company does. That means that you’re going to offer a very small fleet. Even if it’s 3,000, 4,000 or 5,000, it’s still small. A portion of them is never going to be worth it.
If the unit economics don’t work, it doesn’t work.
That means that you can’t depend on the service. That’s at the heart of the problem we’re focusing on.
I need to understand why this is worth pursuing in a city like Boston where it’s going to get some snow. Nobody’s going to ride a scooter for some part of the year. It is the same for a lot of cities. In Seattle, we have scooters, it’s not raining, but it’s wet. I heard there’s something about that, but I don’t know anything. Back up and tell me about Superpedestrian and what the real point of this.
Having done quite a bit of work at MIT on this issue of transportation, what can we do to meet this future transportation demand, which is not slowing down but blowing up? We saw two main avenues. Number one is increased utilization of the car. Instead of having 1.1 people in the car on weekdays, can we get to 4 or 5? We did a lot of work on vehicle dispatch algorithms. Number two changes the vehicles so that it’s smaller. At every point along your trip, you’re using the right scale of the vehicle. If you’re sharing an origin destination with many other people, be on a large view, train view, bus view, or something, but those ones which are small, they don’t exist.
They cost a fortune to maintain. Companies cannot make sense of their business models. They’re not safe enough. Cities can’t manage them. It’s a mess. In these micro vehicles, there is a world where if you addressed technological models, if you create vehicles that can ask themselves if they’re safe to ride before arrive. Vehicles that can predict when something’s about to break before it goes bad to prevent damage from occurring in the first place. Vehicles that can open their own repair tickets and say, “Here’s what I need if they couldn’t protect themselves.” That certainly does not exist, but if you could do that, you’re changing completely the game for a small scale.
It makes sense for fleets, but it doesn’t make sense for auto manufacturers.
There is no reason why it shouldn’t work for auto manufacturers, other than the model of building a car now is very different they have a lot of Tier 1, 2, and 3 suppliers which bringing various black box or technology that you can’t control. It’s a business model. I found it Superpedestrian in 2013, we focused on asking what are these technological bottlenecks that we can unlock, these fundamental technological problems we can solve in order to scale micro vehicles into the large scales that cities meet. We spent the first 4.5 years in R&D. One of the reasons why it’s great to be in New England is you have people that come from the robotic industry specializing in the diagnosis and automated response systems. It is an embedded system control. We hired a whole bunch of smart and interesting people, top engineers from segway before they went bankrupt in the first incarnation who became Amazon Robotics moving the packages in the warehouse.
The VP of engineering for my robot work is cool. A whole bunch of smart people came together to ask, “What is that platform that we can build that will be embedded in the vehicle that can make an affordable, super safe, electric vehicle that could provide the backbone for it.” We call that platform Vehicle Intelligence. We commercialize it in various ways. We have a product called the Copenhagen Wheel, which has a lot of it inside.
Can you describe that real quick?
The key thing we did was we developed a Powertrain. We designed our own motors, choose their own chemistry for battery lithium chemistry, and we build our own battery management system. That’s basics, but that’s not done in micro-mobility at all. For some reason, these are always procured efficient, much safer ways of providing power. We have encryption on the vehicle and User Access Management. We can track the performance of every smart component on the vehicle and also worry about it’s after a sales service in field service on the vehicle itself. If you know how well is my motor controller doing over, communicate that with, “I need to fix it.”
I started to appreciate that. I’ve been obsessed with electric toys so I have been buying a ridiculous number of things I can with wheels and electronics. I never have a clear picture of what’s going on in the controller. A lot of times, there’s some weird incantation you’ve got to do to change a mode. I have an electric dirt bike that you pull the left brake and then the right brake three times then twist the throttle and then turn it off and on again. That’s how to turn off the regen braking. There is all this stuff and I have no idea what’s going on inside that thing. I don’t know if my battery is wearing out over time or what the status of that. There is a lot I don’t know.
I’m expecting it to die at some moment. I don’t have any diagnostics. Sometimes the throttle dies while I’m riding it. I got to restart the thing while moving. I have no way of knowing. I put on three different throttles and it still has this problem. I don’t know what to do. I can see how even for a lot of these products, for them to evolve, that’s maybe not buy another one for my girlfriend or my daughter because it’s a little scary. I don’t mind. I’m fine with things half working, but to be reliable in the way that you’re describing, they don’t even realize that the state-of-the-art isn’t good enough for general use.
The last piece we have on our vehicle platform is decision-making. You can observe it. What components on the vehicle they’re doing and then say, “Can I attribute their performance to failures upstream?” What we realized early on is that the key to be able to do that is to completely swap out all the software of that vehicle. If you want to have full control of the software, not a third party IOT device, third party battery, a third-party motor controller, then you can start to have decision-making systems on your vehicle because you have access. Do we have information that can detect the thing that is about to fail? What fails in the scooter industry? Mostly electronics, otherwise the rest of it is the 1970s, technology mechanical engineering.
It is pretty simple. The L-shape vehicle people abuse its fenders on them yet. The vehicle is very simple. It’s all in the electronics. If you can have full control over the software, onboard the vehicle, now you’re getting somewhere. Now you can respond in nanoseconds, milliseconds, and prevent fires from happening and from going out of equilibrium that if you move away from a thermal issue. If you don’t manage your system thermally well enough for long enough, you’re creating permanent damage to the vehicle. You can put the eventually the rider at risk.
Can I break that down a second? One of the toys I got obsessed with is these electric unicycles, like a solo wheel. I started with the Onewheel, which looks like there’s no way to look cool on a Segway. My daughter looks cool on a Segway, but I can’t look cool on a Segway. I look like the mall cop. I got us one wheel, which is the one you stand on sideways, like a snowboard, and you can look cool and feel pretty cool. It is cool, except that is probably the most dangerous product I’ve ever gotten. I own the fucking lawn darts. That thing is actively trying to kill me. With the Onewheel, you can over torque the motor at full speed because you have so much torque with that deck that there’s no warning. By the time that happens, there’s no recourse. You’re flat on your face at 26 miles an hour. It’s a very powerful and amazing, but the wheel’s not big enough. The diameter doesn’t allow to have enough torque for my body weight.
That is probably the most dangerous product I’ve ever gotten. I own the fucking lawn darts.Tweet
Everybody who has one of these has titanium pins in their elbows now. I’m not kidding. Everybody, if you look on the Facebook group for Onewheel, it’s all photos of guys in a surgery going, “I can’t wait to get back out on my Onewheel,” and their arms are pinned. It’s insane. Fortunately, my Onewheel got stolen. I figured out that those electric unicycles where you have the big wheel between your feet, those things have bigger diameters so they can have higher torque on the motor. You don’t have as much leverage on it. I started looking into it and they’re dramatically safer.
I don’t think they have a higher torque on motor.
You have lower torque as a rider.
Small wheels have more torque. The thing is that because the wheel is bigger, you’re impacted a lot less by the surface of the road.
It wasn’t the road surface. Maybe you’re right.
It felt safer because it’s much bigger.
It’s clearly are safer. That being said, the electric unicycle community seems hell bent on making them less safe by making them faster. Now, you can buy these things that go 45 miles an hour. It’s insane. Mine goes 31 and it’s the first one that has full suspension. I feel safe on it. I specifically want it because Seattle’s roads are shitty and there are a zillion potholes. I could go 30 miles an hour on it. I wear full gear. I have full motorcycle gear. I ride at 30 and it’s got a full suspension. I can hit a speed bump at 20 to 25 and I won’t fall. I wasn’t trying, but I’ve tested that accidentally and I didn’t die. I have not fallen off it yet, whereas when I had the smaller one that I learned on, I wear a full helmet, full motorcycle gear, boots, everything because I know I’m going to bite it at some point.
The point I’m making is those things, the email list or the Facebook group or whatever. For those things is a lot of people saying, they’re posting a video of the thing that went up in flames or they’re posted a video about, “How do I get the settings right to do this or that?” There are weird Chinese devices and nobody has to make their own third party. They’re making apps to try and mod the firmware. Nobody knows what’s going on. Nobody even knows how to change the tire on one of these things. They’re all swapping notes on how to maintain them.
I wish we were at the office because you would see how we do development on each. We have our embedded software team and each person on their desk have three printed circuit boards, five armed processors. That’s the development. There’s no black box. We write code on. We have multiple modularized state machines that work constantly with one another. Your region is an app. Your geofence parking is an app. Your sidewalk detection is an app. Your safe stock, how do you save stock? How do you stop that vehicle if you need to, in a way safe as possible, better, and do its best?
The electric unicycle companies need to come to you guys and get their next-gen firmware.
We don’t sell the platform.
You need to make an electric unicycle, “It’s totally safe and goes 45 miles an hour.” It sounds like you’re ready to do it.
The key to this is the testing that you do. What level of purification you’re looking for before you put the things on the road? We’ve been engineering our control system for the scooters for years. It’s very mature. There are some things we just don’t touch and don’t fix it. There are some things that we need twelve months of regression testing and we’re good. The reason is that we know what our system can do. We also know what it can’t do. We feed some data regarding, for example, we can detect and in real-time, completely resolve more than 50% of issues that would break other scooters into one.
That’s helpful to understand. What’s an example?
A temperature and balance on battery.
When I go to a Bird scooter, and it says the battery is dead, it’s because something like that went wrong.
It was not recharged, there is an issue internally that made its battery damaged, or there is an issue electrically that’s not bringing energy from the batteries. There could be a whole bunch of issues. I can show you the way out. We can log onto a scooter. You can see the log in the back end. There are tons of stuff. There are over 1,300 data points.
A scooter you’d think with two wheels, you couldn’t have that many data points.
We continuously monitor them. First of all, we tried to detect events. We call them events. We have five hierarchies of things that could happen. If you start from info, then we say, “Where is it happening? Is it important? Could it become a safety list? Should we keep an eye on it? There then was a decision-making tree that can upgrade until code red. These are rare cases where we stop arrived. In between, the least and the most urgent is when most of the stuff is going on, where the assistant rebalances itself, thermo management or the system is able to attenuate, energy where we continuously monitor, for example, the integrity of commands and make sure we’re not asking for something reasonable. We have maps on the vehicle.
It knows where it is.
Also, it knows what the city rules are by itself. The good thing about it, one of the biggest problems that have been is that pedestrians end up being penalized. If you want to see tens of thousands of vehicles like this, they can’t be preventing the disabled person from crossing the road or somebody from getting out of their house. Managing them is important.
What city has the most uptake on these things as far as for scooters? Is there a model city?
Model, not really but Paris is doing a lot these days.
What about bicycles? I know some European cities are big on bicycles. Is that part of what inspires you to think that we could have this many of these scooters like Copenhagen or Amsterdam?
Anne Hidalgo, Paris mayor is taking a major action upgrading the infrastructure of the city to be micro vehicle-friendly. They’re investing a lot and they’re taking away car lanes and they’re putting in two wheels small vehicles lane.
Seattle did that. We have them.
During the pandemic, you should see the number of people on bikes and scooters because you didn’t want to drag yourself into the subway. In Paris, traffic is always messy, even worse, and people don’t want to get into a cab or Uber.
They’re using those lanes now. That’s interesting.
During the entire country of the UK, speeding up scooter permitting processes this year. In 2021, just because of the pandemic, there have been places that have been the makers of cycling for a while, like Copenhagen and Amsterdam. Several other cities in Germany and the Northern Islands.
In Copenhagen and Amsterdam, what’s the uptake on like rentable scooters or whatever like?
Amsterdam, it’s not legal. There are no dockless scooters for rent. In Copenhagen, it is growing. It’s pretty good. It’s good, but forget about that, look at what’s happening there with bikes, 50% of the trips, all the trips in the city center. They ride in the winter too.
Will you describe it because it’ll be interesting? What you described is it goes back to your early thesis of saying, “Computers are going everywhere. You guys have been leading the charge on computers that are going to go into scooters.” This is what’s possible. They can be so much better than people even realize. We could be in the future where there’s always a Bird scooter or whatever brand available. Don’t you guys have a brand that you’re working with?
Our brand is LINK. We acquired the company in May an operator of Douglas. We operate in Salt Lake City, Utah, Columbus, Ohio, Knoxville, Tennessee, Fort Pierce, Florida, Rome, Italy, Arlington Virginia, and Provo, Utah. We won to Seattle RFP.
You’ll soon have LINK Scooters in Seattle?
We’re launching LINK in Seattle.
When LINK exists in Seattle, if I go up to a LINK scooter and it’s going to work, instead of going up to a Lime scooter that doesn’t.
When we look at the data we’ve had scooters thrown into the river. I don’t know why people do this, but we’ve had this in Provo and Columbus. We didn’t even design for this. It turns out that this thing can detect that this is happening. We test for watering, but not for submarines. It turns out that these things had an 83% survival rate from deeping into the water. Not only are they recovered. What it does is it allows us to offer the service with the right number of students that cutting the place. Number two is you throw a lot fewer scooters into the environment. I think about it almost a million scooters were thrown into the trash talking about the apparel. Imagine maybe scooters and landfill. It is terrible. Lithium and Aluminum is polluting. We extend the life of these vehicles. We have equal to 2,500 rides and counting batteries that can go on 5 to 6 years. Vehicles since January 2 or 3 times a day are still at 15 cycle counts out of 500 because we have such an efficient powertrain in such a big battery, we can go 50 to 60 miles in a single charge.
Your first product was the Copenhagen Wheel, is that right? Tell me about that. You might be sick of this, but I remember when you guys made the Copenhagen Wheel, I saw it on Weeds. You showed me the Copenhagen Wheel and then it showed up on TV. It is such a good show that was many years ago. For people who don’t know, describe that one because it was pretty cool and pioneering at the time.
We started that in 2008. The question was if you think biking is such a great solution for cities, but sometimes distances that you need to cover is longer, or you don’t want to get to look sweaty. The use case for electric bikes. We started studying electric bikes and we realized that they’re booming in Europe, but they’re very expensive. The median price was around €3,000. They’re expensive. If you’re looking at anywhere mid-end and up, you can easily pay $35000 and they’re okay. For that money we’re like, “You’d expect a lot more. You can even buy a small car for that money,” especially if you want to see this scale into the millions, it can’t be at that price point. We were asking a naive question, which is, “How can you increase their quality so that people can’t resist writing them and make them more affordable.”
At the same time, if you take the whole power, control mechanism and not sell it into the supply chain, if you look at the industry companies, the bike manufacturers buy a motor with a battery and a controller from a supplier, and then they build a bike for it. They sell it to a distributor, to shop, or to you. By the time you get it, you’re paying 3 to 5X on the electronic parts. That’s what drives the high price of electric bikes. We want it to smarter, what if we allowed people to buy direct tiny components? They have only one layer of margin on it. We cannot get a Ferrari for the price of a Toyota. We’ve built this thing. It is a Ferrari, it’s an incredible machine. There’s a culture around it. Many people say the best eBike ever built. It is learning how you move and respond faster than your body can sense under 10 milliseconds.
Is it still the drop-in wheel replacement? How does that work?
The only way to do it so that you can sell direct to power foreign factor was by making it a drop-in wheel. It’s designed in order to be a way to convert your bike necessarily. We’re happy when people do that. You could go to a store and buy any bike you want and make it electric.
Is that happening now? What’s the status of this?
Being this is my first attempted business of tens of millions of dollars, we’ve developed something that’s still an incredible ride. There is a whole cult around it and we make any of the United States. It turns out that if you don’t have a distributor, shops or margin, you don’t have all these layers, you have very little help from the distribution. It means all the customer acquisitions and marketing is on you. With what we put up as fifteen-point margin, more or less, that’s philanthropy. I learned a lot from that. It is an amazing vehicle. It’s the basis of our technology and they are super reliable, more than 55% of the technical issues are addressed without human intervention. On the Copenhagen Wheel, the electrical issues, we know what’s going on. We can detect issues and prevent them from causing further damage. We try to parallel. If you look at most of our work on what I call automated maintenance and self-protection, it’s been done in 2014 and 2015. There is a lot of pens and sharing and did a lot of developing this platform. We tried to sell it initially to other operators and tried to sell it with the few large ride-hailing companies in 2016 and 2017, to the large bike-sharing operators in 2016 and 2015. There was very little desire to go electric.
Basically, what I learned are two things. Number one is you don’t want to take everything on you. Reinvent the marketing channel, the sales channel, and do all the technology and all the manufacturing and all the supply chain and all the purchasing. Everything is new. It almost no off-shelf part idea. On the other hand, what we did learn is that technology that we’ve made works and real. It’s fundamental. This issue of detecting problems in real-time like water, temperature, balances, thermal management, and that has many applications. We’re monetizing this now in the scooter world because I’m passionate about its mobility. Scooters, the first mode will be the other mode. We will be enabled in many ways by reliability, safety and scale. That’s a way to completely drop the cost of offering a service is different using this technology. It’s about 75% cheaper who offers scooter trip with this platform. It changes the game. That’s an example of how you can start to embed. The intelligence could also be local in the thing. The users don’t even know about it, to seize your scooter, you unlock it, ask yourself if it’s safe before every ride it tells you this is safe, but otherwise, it should be a street bench on wheels. Be there when you need it. That’s different from the high-level analysis or the centralized Oracle view that you were talking about back to the beginning of our conversation saying, “How to disrupt the contagion chain in the pandemic,” which is a different way to think about what computers aid us with. Think of it locally.
It’s an example where you’ve trained the device to know what it’s optimal operating circumstances and what to do in response to different failure modes. Your knowledge keeps improving. Does the device itself keep improving? Is it connected? It’s online so it can talk back to you. You can see the whole fleet.
We started from an embedded system, but it’s all online, the cloud. That is in conversation with embedded soft.
I love that because my scooter’s learning from all the other scooters. I think the public discourse around these devices is missing this key attribute with robots. These devices, they all learn from each other. When one of them gets thrown in the river and figures out and you figure out what to do about it, they all learn what to do next time. When a Tesla, most somebody down, walking in a dark alley where they’re not supposed to be crossing the street, all the other Tesla’s learn never to do that again. That’s a terrible situation, but that’s not true for human drivers when a human driver most somebody down, then no other human learns from that. You seem more bullish on human intelligence and decision-making than I am.
Didn’t you teach me that analog visual hacking is easier than computer-based stuff?
It’s true. There’s an opportunity here to see these technologies as being able to help us do a better job. You’re proving it in a sense because a scooter is like another place where a computer hadn’t gone, now it’s gone there and we can use it to make it a lot better. By extension, I’m sure you got the fundamentals of what you guys are building could be used for any electric vehicle, if not other electric stuff.
Our platform now, the system is directly applicable to any vehicle under 12 kilowatt per hour, so the battery to motors.
Is that anything DC?
We were in a 52-volt DC system, but it’s just a circumstance of the power rating of the different components, but if somebody wanted, they could put this in winter for detection or prediction of issues.
That is a big deal. That’s exciting. What do you guys need for Superpedestrian? What would be helpful for you? Do you guys need more scooter deployment partners? Do you need windmill companies or like what would be helpful to make you guys turbo boost the company?
I didn’t come with an ask but I will tell you where we are. We acquired a company in May and since COVID lockdown to remove, we expanded fast. We entered Europe and hired some great people. We have this special engineering team that’s been built over seven years. That’s great. We then acquired a company that’s ten years old from Baxter. We acquired their fleet management business. We closed the gap and then some on operations on the ground. We have our own W-2. We have our own employees, but then we needed to close the gap on because of the government relations side of things. How do we directly the city? We hired some of the best people from across the industry and senior executives from other scooter operators. We are happy about that. We also fundraise some more. We’re good on that front as well. What kind of help? Most interestingly, for us, is to get as many opportunities as possible to show cities our ways of offering safe, manageable micro mobilities.
Cities are tenuous because they’ve had some mediocre operation come in, make a mess, end up with a lot of SKUs in a river.
A lot of promises were made by essence operations company but with engineering capability. There are a lot of problems, but many of them were not catched.
The cities got burned. You want to be able to show them like, “We can do better. We can be 75% more efficient. We can come in and make these things safe and reliable.”
We look forward to a less affluent community. It could be a lot more dependable because it’s there. You need to manage it so that it doesn’t ride in a park or next to a school. Our vehicle stopped within 0.83 seconds. We’re at 99.91% success because the maps are on board, vehicles autonomously enforce the rules. It’s very different. It’s like half a minute without a vehicle. If there’s no connectivity, sometimes it’s never enforced. There are 250 cities around the world that didn’t have scooter licenses. The number is growing to over 600 in 2021. There are more cities to deploy scooters. There is supply even throughout the world, irrespective of a lie. However, it’s important to tell the story that technology and capabilities matter.
Cities need to know that they don’t want to go make the same mistake and end up getting burned in the same way.
Otherwise, it is just a commodity.
I remember seeing piles and piles of rental eBikes or something. In Beijing, companies tried to deploy right. Bikes for rent and that ended up failing somehow. There are stacks of them.
That’s an industry where the unit economics works.
It baffles me how these things go. I want to let you get out of the wind here, but let me ask about one thing because this is interesting to me. One of the projects I remember you doing years ago which I guess must have been in the Senseable Lab was throwing away GPS trackers and the trash maybe all over America and then seeing where the trash goes around the world. This is an epic project that everyone should see even now. It a decade or more ago, but that was an amazing enlightening project where I could see for the first time what was invisible to me. It was so much more dramatic than what you would have guessed or what I would have expected.
If you ask me at that time like, “What happens when you throw a can of Coke or a pair of sneakers in the trash,” I would have thought, “It gets picked up by the truck that goes to the transfer station where they send the recycling one direction and they send the trash another direction. The sneakers go to a landfill that’s 30 miles away and the can goes to a recycling center somewhere in the state. It gets becomes aluminum that’s melted down and turns into cans again.” That’s what I would have guessed. That is not at all what you showed. Better than me trying to do a shitty job of summarizing, why don’t you describe what that was about?
The genesis was in 2008. We started seeing anecdotes about how waste moves. We wanted to try to verify this with a quantitative study. How do you do that? If you’re a waste management operator, all you have is a statistical data. You don’t get to know the pieces of trash flow through your system. We built these sensors that are IOT devices. They detect their own location and they communicate. We partnered with Qualcomm. They need to survive in the trash. That’s quite an IOT device. I was talking 2008, 2009.
Was it easy then?
It wasn’t too difficult, but we’re not as good. We wanted to get six months of life of those. That required a lot of smarts and how you designed to power. The key thing there was we went to Seattle on purpose because it was their recycling time squad in the United States. Most investments, a lot of compliance on behalf of citizens. We wanted to see how it functions. We partnered with waste management, one of the biggest waste removal company, maybe the biggest back then, Qualcomm, Architectural League of New York and the City Hall in Seattle. We then invited 500 households. We gave them a list of average household goods that we want to track. We went to their home to help them tag their trash and told them, “Throw it away as if we were never there in the first place.” There are various other programs and that’s when people brought it over to the library. Long story short, the first deployment was a few thousand pieces of trash that people put away. We started getting in real time because what we saw is that the things that we learned to recycle in the previous century, and been doing for a long time, metal, glass, and paper, have been getting to their end destination quite effectively.
In three days, they get processed. We saw some anomalies, like some cheating, some recycling going into it, getting buried so you can catch some fraud. The interesting thing that we saw is that the newer thing to recycle, like electronic waste, nobody had a full bird’s eye picture of it. Let’s call it the removal version. We saw things going from Seattle to the East Coast and then back to the West Coast for processing. There’s a tipping point where the emissions due to transportation, the advantages of recycling cell phones with the Florida printer cards went through Southern California. There are a whole bunch of things that we started observing. There was then a big flow to the Nikon Delta, which we did not yet understand. That led to a follow on project with an undercover operation that ended up exposing the flow of CRT (cathode-ray tubes) monitors to the Mekong Delta instead of being recycled in California. Governments would pay US companies to recycle CRTs because they are toxic.
Because they contain cobalt?
We need to dispose of them through particularly good control process and expenses. Instead of paying for that process, some of these companies took the money and ended up shipping them to be reinstalled in products or another life of these DVDs combined with a monitor repackaged into new cases. We put cameras as well as backers and these CRT monitors documented their road trip.
Like GoPros in them or something like that and they’re just recording.
It is based on Android. There were multiple incarnations of the same product, which as you said, allowed us to take something that was fairly invisible before sprinkle a little bit of digital stuff on it, to bring it to light and learn several things
I assume that I can find that stuff online still.
The original product was called Trash | Track.
I don’t know if you’ve followed this, but in the last few years, we’ve been able to reflect on recycling. Some important studies have been showing that we probably shouldn’t have done it. We’ve been doing recycling in the US for 40, 50 years. There are almost no cases where it’s working at all even now, which is sad. Especially, living in Seattle where there’s a lot of compliance. Everybody follows the rules, “We all separate our trash. We all do our part.” We all feel like we’re solving a problem and we’re not. I feel strongly that we’re running out of time to do these feel-good things. We need to be focused on metrics and figuring out what’s working and what’s the potential to work.
With recycling it’s, I think fundamentally, it’s an energy issue. If you have cheap carbon-free energy, then you can recycle stuff. If you don’t, then you’re going to burn more coal to recycle stuff than not. I don’t know if you saw this. I have to look this up to show you, but there’s a pretty damning analysis of recycling for plastic, less than 10% of the recycled plastics we get are significantly reused. They’re trying to frame it as a case where the plastics industry tried to promote recycling as a way of making people feel like it was okay to use plastic.
The whole premise of plastic to begin with is to do something indestructible.
It’s great for that. At least to last as long as I do and that’s good enough.
George Carlin used to say, “There is a reason why God made us the first place.”
I never saw that. I got to find it. I haven’t gotten to spend a lot of time on it yet, but what I think is probably the right thing to do now is go evaluate what cases we’re recycling works has near-term potential, focus on those, probably get rid of and shut down the rest of them, and then focus on improving energy. If we can make abundant carbon-free energy, then we can get back to recycling.
It’s a very good point you’re making. There is another world which is the material world. When you blend materials together, sometimes, there is no amount of energy that will get them back apart, which could be crucial for recycling. Most part, with clean much cheaper energy, we’ll be doing a lot better. Your point is very important. We saw a similar thing with electronic waste. People would come with a laptop and pay $20 in 2008. In Seattle, they would pay $40 to get it recycled. They didn’t know that this thing would travel 6,000 miles.
There is exactly. In Seattle, everybody’s going to follow the rules and do it. It’s insanity. They wouldn’t know either.
They had nine links on the way from the person who drops off their trash until the final facility. Sometimes nine service providers, each handing a part off to one or another,
None of them realizing the origin or the destination.
“I shipped it to Florida. I shipped it to California. I just shipped it to Chicago. It’s the battery and it’s the plastics now. It turns out this whole thing travels all around the country.
That’s an awesome work. I’m excited about that project. It made an impression on me. Hopefully, other people got that. Hopefully, there’s still some life in it and we can spread the word about it some too. We can wrap up and get out of the wind. We have some pretty interesting audience of people who are excited about technology. I think of them as a pretty high-value audience, so if there’s anything you want to share, we can do that too.
I shared so much. I’m excited now. There’s a moment where robotics can take so many shapes. I focus on the robotics inside these rentable vehicles allows us to make them safer, more affordable, more ubiquitous. There have been like 30, 40 years of work since the late 60s about robotics. It’s come to such a point where we can start to deploy that incredible technology in ways that make people live a better life, better with the environment, and make us be more healthy, productive, or having more fun. Transportation is my area, but I think it’s an exciting moment because we’ve been talking about what to do about transportation problems since the 70s. It is the first time where we see something fundamentally new that people love and cities love as well. That works technologically that can be scaled. We’re at the beginning of something. It’s not the end of all the hype. We’re scratching the surface.
I had this experience a couple of years ago. I was in Brooklyn with a buddy of mine and I was going to hang out with him. He was at the City Advisory Council meeting, where I went to meet up with him. I sat in on a little of this meeting and the meeting was about what to do for the future of transportation in Brooklyn. I want to say years ago, I wasn’t invited. I’m just sitting there. These guys gave a presentation about their vision for the future of transportation in Brooklyn. Their idea was to deploy streetcars in Brooklyn. Now, Brooklyn doesn’t have streetcars, but it did a hundred years ago. They tore them out.
You still see the tracks.
In a hundred years, the idea has progressed from streetcars to streetcars. It drove me insane to hear them talk about this. They were talking about the economics of it. Apparently, it’s like $5 million to build a streetcar. I literally grabbed a napkin, started writing down, “If a Tesla costs say $50,000, how many of them can I get for $5 million?” They’re going to tear out road lanes and put in dedicated lanes for streetcars. I’m like, “How about if we make those dedicated Tesla lanes and we buy a zillion Teslas and we programmed them to drive around and pick people up and drop them off? They can do that now in Brooklyn. If they had their own lane, they could certainly do it.
Doing it is a lot cheaper.
You could do it cheaper. You don’t even need a Tesla.
You can create an electric bus.
It’s insanity to me that after a hundred years, the best idea Brooklyn can come up with is streetcars. I got kicked out of that meeting. I haven’t been back to Brooklyn since. That’s the thing is we’re not doing a good job of asking ourselves, “If we were starting from scratch, what’s the best that we could do?” Use that as our metric for the goal and say, “With no new breakthroughs, what’s the best we could do if we were starting from scratch?” I know we got to knock that down a little bit because of tradition and regulatory issues and maybe some safety things we hadn’t thought of and whatever people are used to. You’re not shut down a little bit from optimal, but we could do so much better. People are not thinking that way and they set their sights to low.
The thing is that there’s so much understanding on what can be done before you have been introduced technological innovation to improve on urban mobility. Urban mobility starts from how you plan a city. If you live above an office, that’s above a restaurant, you’re going to drive a lot less. That’s been well figured out since Jane Jacobs years ago. There’s the world of planning, but cities changed slowly. It is not always under the ultimate control of the city because there are many pressures that the city has to deal with. Now, as an example, cities had to spend money on the virus. Some of the programs have to go out the window to find new ways of doing a thing.
If you were going to describe urban mobility in 2030, how do you imagine a city? You could pick any city in the world and say, “Here’s how good it could be knowing what we know now.”
I think we can fantasize about it to get it. First of all, we know is that most likely cities are going to look pretty similar. There’s a good reason. The lowest energy is required for us to move on the surface. Elevating us makes no sense. This whole idea of all that stuff to me is baloney. It can be done, but your question is, why would you ever want to do it if you’re going to spend an order of magnitude more?
It’s so much energy, it’s noisy and bulky.
What’s the point? Maybe for wealthy people to go from SFO to Palo Alto, that’s great, but that’s not the killer app.
The killer app is to raise venture money though.
I respect that because maybe one day we’ll become ubiquitous enough to replace aviation with something electric. It doesn’t solve over urban mobility. Urban mobility is going to be a combination of mass transit from subways to buses to electric buses to call corralled electric vehicles. Whether on a predetermined route or on an agile flexible route mass media. You and I shared the same origin and the same destination along with many other people in combination with a multitude of other things. Other modes of walking, scooters, tiny little electric cars, like coats, and the combination of those, I think the common denominator is going to be slower and more individualized.
I call them packetized. If you look back at internet architecture, we used to have this like hub and spoke thing where it’s like the telephone of the 70s. Every wire from every house went to a central switchboard and you had to have a wire from there all the way out to every house. That got replaced with TCP IP, which is packet switch, where you got to get near something that’s online. The packets, can be routed to their destination. That’s what I see is wrong with a lot of urban transport is that it’s not packetized. You get on the bus where you at the bus stop, you get off at another bus stop, and instead of taking me like Uber does from where I’m at to where I’m going. It takes me all the way to where I’m going. That’s got to be part of what makes it more successful because you reduce transitions and you make it. You increase convenience, reliability, safety, and all those things.
Let me give you an example. We’ve looked at 45 cities around the world. The demand for transportation in them, this is through various publications at Senseable City Lab to do that at MIT over the better part of twelve years. Most of that data we got from cell phone service aggregated or anonymized, or a combination of those tell mode independent mobility, like walking included as well as cars must, as well as the subway. How do people move at large scale? You take that data and then you superimpose it on the city’s transit system. You see exactly what’s missing because you say, “I want to get from my origin to my destination.” You wouldn’t care jumping on and off various things if they seamlessly connected with it. That’s going to be fine, especially if it’s affordable and reliable. If you save an hour in the process, we overlaid that. We see, “First of all, which types of vehicles are missing?” You can see that there are some trips that are 1 to 2 miles, and then you got the 2 to 4 miles, and then the 3 to 7 miles. In the vehicle categories, you can see how many people shared routes, given time along same parts and/or you can decide if is it a 1, 2, 3-person vehicle that I need?
Google Maps could tell you all these things now in real-time.
Google maps, if you chunk it into trips, that’s relatively simple logic. We had to deal with the same problem. We’ve done a lot of thinking about vehicle form factors. How big do they need to be? How many people did they need to take to be able to help you address as mobility demand as possible? We’re starting with this tiny and basic thing.
If those things existed, like the Uber app could ask you how many passengers and send you the right car or vehicle for the job.
You can easily imagine it. There’ve been in this future that you get from your home. Let’s imagine you live in a suburb. You get from your home in a one-person car or on a moped or something like that. You dropped at a transit hub.
An Uber could come by and drop a scooter off in front of your house if that’s what you needed at the moment. The drone could deliver a scooter.
Those things are becoming so affordable. You can also overcompensate for them with greater numbers so you wouldn’t need to drop them off in somebody. It’s a matter of lowering utilization.
Thanks a ton for taking the time to do this with me.
- Institute for Disease Modeling
- Copenhagen Wheel
- Architectural League of New York
- Trash | Track
About Assaf Biderman
Assaf Biderman is an entrepreneur, author, and technology inventor. He is the founder and CEO of Superpedestrian, a robotics company that develops platforms of small electric vehicles for shared use. Together with the team at Superpedestrian, Assaf has developed fleets of scooters, e-bikes and other micro-vehicles with autonomous-maintenance capabilities and active-safety systems that enable much safer, cost-effective shared mobility services.
Assaf is also the Associate Director and founding member of the MIT Senseable City Lab, a research group which develops technologies in big data, machine learning and robotics aimed at improving livability in cities. He has supervised research in areas of urban sensing, data fusion, and urban transportation, and also leads lab partnership initiatives with cities and the private sector. Assaf has a background in physics and design. He holds over 150 patents and publications, and has been honored with multiple international awards including the Red Dot Luminary, Time Magazine, Thomas Edison, and James Dyson awards.