Prototyping and Testing Smart City Innovations

Speaker: Kent Larson, Director, Changing Places, MIT Media Laboratory and MIT Living Labs

Transcript

Gordon Feller

We have an opportunity to cast our eyes, South Africa, North America. Now Kent Larson is going to take us to some other places and these are stories that we’ll have a chance to talk about tonight over dinner and to continue the conversation tomorrow. Kent Larson has been a practicing architect; 15 years in New York doing that. He not only works as the director of the Changing Places program inside the media lab, but also is responsible for the lab that you don’t know at MIT, which he’ll tell you about, and which is called the Living Labs. He has a practice from MIT that really is global and we’re really pleased that you could be here with us, Kent. Thanks so much for joining us and we’re looking forward to your presentation.

Kent Larson

Where is the little slide? Okay, thank you. Could I have the slides, please? Great. Well, it’s great to be here. I’ve met some great minds and made some new friends. Thanks to Gordon for inviting me. You run a tight ship, I appreciate that. I’m an architect. How many architects are in the room here?

There’s a few. Okay. Let me start with my beginnings. This the first major project I did pretty much right out of school. It was in addition to a [inaudible 01:34] house, one of only a few in the U.S. You can see a little bit of the house on the right. We did this pavilion and pool adjacent to it. It still has some of the ideas we eagerly play with. The doors go into pockets that can be all open, all glassed, all screened. Glass and curtains, there’s a lot of prefabrication. This was done over 20 years ago.

The problem is this doesn’t scale. This was a unique solution to the needs and values of the rich client. That doesn’t scale to this problem, extreme urbanization. Right now these two lines are crossing. The red line is urban population and that’s accelerating rapidly. In fact, you probably all know, the next 15 years or so over 300 million Chinese will move to the city. That’s equivalent to the entire population of the US, so we have to build mobility systems and housing to accommodate that, and we’re not doing too well.

This is the kind of cities that are getting built. They’re basically generic commodity, single-purpose ghettos that are reliant on private automobiles. This is a satellite image that I found. You can see the kind of cities that are being created. These don’t have the attributes of the great cities that we all love.

We’re not doing too well in the US either. This is Huston. Every little red square there is a parking lot or a parking deck, and so our cities are getting eaten up by this, and we have to do better. In fact, one study showed that in congested areas 40% of the gasoline is used just looking for a parking space, and transportation and buildings that count for something like 60% of urban energy use. We’re trying to think about buildings and automobiles.

I took this video in Bangalore a few months ago. A friend of mine from Bangalore said 15 years ago these streets were filled with bicycles. You don’t see any bikes now. In fact, what you see are mostly private automobiles. Here’s the people. These streets are not designed for people, it’s not sustainable. This is Korea. You can see the kind of streets that we’re creating.

I’m from the Media Lab, as Gordon said. We all have to write a how statement, and my statement, from my group, is how new strategies for architectural design, mobility systems and networked intelligence can make possible dynamic, evolving places that respond to the complexity of life. Basically, we start with people, what their needs are. We try to deploy a good design and then we add technology when it’s useful, and we try to find interesting problems to work on.

I’m going to talk about four scales of problems that we’re thinking about. First, New Urban Strategies. Basically, from our point of view, that means minimizing the need for mobility, creating agile spaces, ending zoning by use ñ office buildings are dinosaurs, we shouldn’t build them anymore ñ creating some unique identity. So a 20-second history of Chinese architecture. You have the traditional urban fabric that was basically before 1920. The grid introduced in the 20’s, you had more soviet-style blocks in the 80’s, and now you have these high-rise superblocks. That’s only interesting I think in this respect, in the context of this meeting.

When you look at the energy use, on the right you can see the superblocks, the top squares, embodied energy ñ it’s about the same in all four building forms. Mobility is double or triple. That’s because it’s reliant on private automobiles and the building use is double roughly what these other forms are. There are a lot of reasons for that. The space is bigger; they have more air conditioning, more glass.

We have an opportunity now in our group to test some of our ideas that I’ll talk about in a minute in two new towns that are just to the west of Shanghai. It’s a consortium of Finnish companies ñ you can see the companies here ñ that are putting 60% of the equity into two new towns, and they’ve asked us to help them on the research side. They have a pretty interesting agenda. I think they, for the most part, got the big things right. It has a terrible name, DigiEco City, but it means incorporating digital concepts, ecological concepts and city concepts. These are the ones that we’re particularly interested in.

We’re trying to get some traction on this idea that if we take the density of these superblocks and we combine it with the walkable, traditional form of the traditional cities, then you can have the density with the walkability and reduce energy, make more livable cities and generally make better places to be. Basically it’s taking whatever traditional streetscape. It could be a modernist streetscape, but places for people and hiding in the back of them, the functional spaces, the places where work takes place and where people live.

You can find precedents for this. Rappongi Hills in Tokyo is a great example. This is an aerial shot. It’s evolved over the years. These towers are the efficiency layers – that’s where work and living takes place – and then you have the experience layer, at the streetscape. They really have separated those two in quite a beautiful way. It’s a very complex environment that has many of the amenities that we think a great city should have. We’re trying to tease apart pieces of that and develop scalable strategies and methodology that we can take to some of these new cities, like the mobility spaces.

First of all, we’re starting with a typology of what we call pathways or streetscapes. A couple that I want to focus on are the people, spaces and what we think of now as mobility parks. I had an opportunity to be in Melbourne, Australia for two weeks this summer and I loved the laneways. How many people know Melbourne? It’s one of the great cities of the world. They had these alleys filled with garbage that gradually they’ve been reclaiming. You can see this is a highly evolved laneway. There are no vehicles, there are no bicycles. It really is a beautiful people space. They have cities from all over the world coming to Melbourne, asking them how can they duplicate the laneways.

This is what we think of as the mobility park. I shot this here, in Boulder, yesterday. You can see the bicyclers go under the bridge, the car goes over, the traffic’s separated. Here’s a little pedestrian path. You run or you bike along the river. It really is a terrific space. Many cities are now thinking about how to incorporate some of these ideas, but I think Boulder’s done a beautiful job.

I shot this last week in Seoul, Korea. It’s their equivalent of our Big Dig. It’s another linear park and it is beautifully executed. It has places to play and to meditate and to read and to work. It has transitions from modernist spaces to natural spaces and it’s a pathway. Up above the activities of the city go on and they’re separated. This is a beautiful refuge that’s also in effect a mobility park. In this case they don’t let bicycles in. But it could be.

This is our equivalent. It’s one of the great disappointments, I think, of urban planning. I think of it as the Rose Kennedy traffic island. Anyway, it’s not what we should be doing.

Let’s talk about now the mobility side of that. So with this we try to come up with concepts that will allow a population to be served with far fewer vehicles that use last energy, less land for parking. All these things you want to do in the city. Now there’s a trend to this direction. You all probably know about this, but there’s congestion pricing in London. Beijing has limited new vehicles to 240,000 a year. There’s bidding system, taxation to reduce private automobiles, and then you have the emergence of all kinds of vehicle sharing, certainly bike sharing. Boulder has one, Boston just got one last month. Car sharing that’s more fleet-operated, like ZipCar. We now have peer-to-peer car sharing. This certainly is a trend.

Our idea is that you have this cloud-based mobility on demand, which is an ecosystem of shared vehicles connected to mass transit that allows a person to use the vehicle at the time and the place that they need it and use it in one direction and then can switch modes seamlessly the data about location and demand profiles, etc., all goes up the cloud and it’s all accessed through personal devices. Chris Borroni-Bird talked about their concept, very much the same thing that we’re thinking about.

We like to build stuff at the Media Lab. This is a city car project that we did. Now if you look closely, some interesting things are happening. Each wheel is controlled independently. It can spin on a dime. It can go nose in to a parallel parking space. The car folds, so the length is the width of an automobile. The front door opens, you step directly out on the curb. This is the kind of vehicle that we’re developing. This is more optimized for urban use. This is Ryan Chin and me testing this ergonomics model to look at the viability of folding. And this is the unveiling of the vehicle in Spain a few months ago.

Just to give you a little sense of what this vehicle is like. The project was started by Bill Mitchell. I inherited it. Many of you probably know Bill. Unfortunately, Bill died just as we signed a research agreement to work with a company to commercialize this. You can see the length of the vehicle in the lower right compared to a smart car. It’s much lighter. We’re actually now working with AARP to look at concepts integrating personal mobility like wheelchairs, directly into the vehicle so there could be a dock; a very interesting concept that we’re developing.

The key to this vehicle is the robot wheel. You have the drive motor, the steering, the suspension, the braking, all integrated into this prefabricated package that bolts onto a blank chassis much like you connect a USB device. I wanted to mention this is the [inaudible 13:20] model because the [inaudible 13:22] people are here.

This is a consortium of companies that are involved in commercializing this. It’s in the Basque region of Spain. It’s companies that are suppliers. There’s no OEM involved. This is an experiment in distributed manufacturing. These are the components that each of these co-manufacturers – I mean, we don’t call them suppliers, they’re co-manufacturers. They make them independently and they’re coming together to form this vehicle. The idea is it’s all mass customizable. This is one version of many different vehicles. We could have taxis and dump trucks and whatever we want by using the same basic infrastructure of the robot wheels, would drive by wire electronics. It’s an electronic vehicle that is, like I say, optimized in this case for city use, but it could be delivery in India.

This is the mechanical platform, as it existed a couple of months ago. You see the battery pack being connected to the electronics. This is the production robot wheel. This is Will Lark, my PhD student in Spain, with the first full scale body prototype. These will all come together in the next couple of months, so we hope to have a fully functional prototype of this operating at some time end of October.

The interior has a yoke. There’s no foot pedals. You have to step directly out. It’s like an airplane yoke. It pivots, so it can be on the right or the left. All the controls are by hand. You can operate it in the US or England. We’ve had fun experimenting with different interfaces that sit on that little display in the middle of the wheel.

Anyway, that project’s finished, at least our part of it. We’ve done a hand off to industry. We’re now looking at things like autonomous parking. You heard about that this morning in the first session. This is our little half-scale prototype at the Media Lab running around in the lobby. We’re looking at strategies.

First of all, for autonomous parking, it’s pretty easy to do. There’re no technical barriers. It’s very different than the Google car because we can introduce infrastructure into the building that communicates with the vehicle, so you can greatly simplify the technology that’s needed for both. Basically you pull up to a drop-off point, you pat the car on the butt, in it goes and it parks itself. That’s our model. With Chris’s car, you zap it with a remote. Anyway, we’ll work on the interface.

Now it gets very interesting, this idea of combining folding and autonomy. If you look at the cost of a parking space in a underground parking deck like we would build in Boston, it’s about $1,000 a foot. Sorry, about $80,000 per car. If you can combine autonomy and folding, you get about an 8 to 1 ratio in that same parking deck. You can actually do better if you optimize the design of the parking. So that brings a cost per car down to $10,000. You save $70,000 per car by combining autonomy and folding. A developer could give away a car, put money in their pocket if the city would allow them to build less space for parking, a very interesting concept.

We’re now working on this vehicle. We call it a persuasive electric vehicle. We think of it as the missing link in urban mobility. It’s democratizing access to bike lanes and addressing the problems of energy, congestion, mobility, ageing and obesity simultaneously. Basically, what we’re trying to do is to decouple the drive from the human input, developing an interface that encourages people to get more exercise. If you’re an elderly person it’s basically a switch and if you’re an athlete, you can put more energy into the system that you use and go forever. So you pedal to turn a generator that charges the battery that drives the motor.

This vehicle also folds, so we can get two of these in the length of a conventional bicycle. This is the little animation study. The design is not quite right. This is our first pass at it. We will go through two or three more iterations this fall, but you can see how it might operate on a bike lane. We’ve talked to some of the great bike cities, like Copenhagen, about this. They’re very interested. They have a lot of three wheelers in Copenhagen, by the way.

One of the problems with this shared use system is they get out of balance. So this is how they rebalance them for bike sharing. They put them on trucks and they move them around. It’s very expensive and it doesn’t work for cars. We’re looking at interfaces that automatically will redistribute those vehicles by creating incentives to do that.

Sorry, this animation started a little quickly, but let me catch up on that. What that was were mobile phones moving through the city and the network here was looking at where people started, their destination, and where they ended up. And they classified them according to, in this case, nightlife tribes. It looked that nightlife behavior of people ñ young males going to the clubs or young families going to Denny’s. It clusters them by color and then remaps them on the city. Just as an example of the kind of data analysis tools we can use to achieve a very rich understanding of human behavior in the city.

The third scale problem we’re looking at is what we think of as responsive places of live-work. I separate live-work because I think we need to. It’s the same idea as the car, basically ñ use space more efficiently, consume less energy, and respond to personal needs and values.

One thing that we’re very aware of is the nature of work is changing. Companies are becoming virtual. At any given time many companies say a third of their workforce is in the office, a third is in third places like Starbucks ñ you can see this is basically a workplace ñ and a third is working at home. That makes the home as important a workplace as the workplace and we don’t design workplaces and homes to acknowledge this.

The other trend we’re very aware of is personalization. Essentially every consumer product now is personalized in some fashion, if you choose to. There’s thousands of mass customization configurators for everything from blue jeans to wedding rings, certainly shoes. We don’t do this with architecture. This is a photograph I took in Taipei. That little box in yellow is the only part of the building that remains as the architect and builder intended. Everything else has been customized in some ad-hoc way and it’s terrible design but it’s a beautiful expression of this desire for personalization that we haven’t figured it out how to respond to in an industrialized way.

What we’re doing is, much like the car, we’re standardizing the chassis and keeping that unencumbered. There’s a Finnish company that we’re working with, it makes galvanized honeycomb panels, the same material use for walls, floors and ceilings, stacked them very efficiently, very lightweight, very energy-efficient. Our process is you build these chassis, the upper left is what the builder does, but then there’s a personalization process to respond to needs and values. The way we think of this process is understanding who you are. We use image sorting, we use sensor data, we use Facebook data, certainly questionnaires ñ lots of different ways of doing that. If we understand your profile and we have profiles of the architectural solutions in a fine-grain way, we can run them through a matching algorithm and they give you a ranked order set of designs.

This is not architecture as I was taught architecture, but it’s scalable architecture where the skill and the design is in the system. It’s more of an industrial design process. And then we give people a refinement tool. Can you turn the sound down of that? Sorry.

This is a little interface we built where here the person interested in housing is reading articles on the New York Time. They see a link to a new service. We’re actually talking with the AARP about doing this with new tablets. They’re sorting images. It understands their profile. It assembles a unique design and based on what it thinks as a best fit to their profile and then it gives people a refinement tool where they can go through and drill down into that design as they like. It’s a little bit how an architect works with the client. You don’t ask people to place doors and windows, you show them a design and they respond to it. We’re trying to simulate some of that process. Not just the design, but the services and the technologies that can be integrated into it.

Now one of the particular homes we’re most interested in, we call it the City Home, which has a very small footprint at least by US standards, 800 sq ft, that transforms from an exercise space to a work space when your employees come in, to two guest rooms for sleeping, to everyday use, hanging out with your wife, to a space for a dinner party or a larger cocktail party, if you need the extra space or maybe even it’s a dance studio.

The challenge here was how can you make an 800 sq ft apartment function as if it was 1600 sq ft? Assuming that all of those functions that it can accommodate were single-purpose, it would be the equivalent of 1600 sq ft. The architects have been playing with these ideas for a long time, but we’re interested in trying to find industrial processes that can allow this to happen in a very economical way.

These two projects come together in the City Home. We’re looking now at even more efficient ways of storing vehicles in this context. Here you see the City Car dispenser with our personal electronic vehicles and our City Cars in this little dispenser. Here you see a City Car approaching the City Home and beyond. There’s a car being lifted up into this dispenser. It has a little undercarriage connector and a robotic charger. It’s all charged automatically. I think this idea that you charge your car like you fill up a tank with gas is kind of silly idea. It’s sort of horse and buggy thinking, but we’ll come up with better ways.

One of the projects – also that I should mention in this context – is the one we’re doing with Schneider Electric, and that is looking at repurposing automotive batteries that have reached the end of their useful life into a buffer that can be located in a building that’s charged when the loads on the building allow it, and then rapid charging with a direct level 3 DC to DC charging. By the way, it can go the other direction, too. In a peak load situation it can go back up to the grid in an emergency situation. We’re actually building one of these stations on campus in partnership with Boston Power and Schneider Electric and I think this combined with fleet operation could be the answer to the storage of electricity in the city.

The last scale I will now talk about is what we think of as proactive personal technology, and this has two components to it. One is automation systems that respond to human activity and the second is persuasion, persuasive interfaces to help motivate a behavior change. If we go back to this diagram that I had of Chinese cities, that big blue box is what we have to address and much of that is energy consumption related to day-to-day decisions of people. You can’t deal with it without addressing behavior change.

So what’s a persuasive interface? This is a good example; just about everybody hits the brakes. It’s information at the point of decision ñ quite persuasive. This is a little more sophisticated version of that that we developed at the Kendall Square T stop. You go through the turn stop, you have a decision to make. You go to left to take the stair, the right to take the escalator. We got them to allow us to put a box with a computer vision system and a projector. We automatically counted the number of people that did both baseline data over many months and then we projected this intervention at the point of decision. Your heart needs exercise, here’s your chance, so you can see a tired sick heart riding the escalator and then a healthy heart leaping up the stairs. We can show thousands and thousands of data points that with the right message at the right time and the right place we could change behavior.

Then the question is how can we do that in the context of a personal environment? This is a living laboratory that we built some years ago. We call it the Place Lab. It was a prototype in that it was built from these prefabricated cabinetry components with integrated technology that we advocate for this mass customized housing, but it wasn’t an apartment, it was a living laboratory where volunteers would live for some weeks or months and trade away their privacy in the interest of science to test some intervention that we would then put in front of them. Behind the scenes were hundreds of sensors that allowed us to know where people were, what they were doing, the objects and the systems they interacted with and media that could communicate with them in the context of that activity. We also set as one of our goals to make this facility obsolete in three years ñ we had an NSF grant to develop tiny little wireless sensors that we could sprinkle around any environment. These are the different variations of those sensors that we call MITES ñ MIT Environmental Sensors. We’re not deploying these.

We don’t need the Place Lab anymore. We can instrument any environment. We can put them on in this case a spice rack or under chairs. It’s a little two-access accelerometer that wakes up and broadcasts the signal. We have a three-access accelerometer you can wear on your body, so you can see hip, wrist, arm, ankle, thigh ñ that first risk motion can be revealed by the acceleration signal. We can then get fairly accurate activity recognition rates.

We’re now actually doing this directly to a phone. We don’t need a computer. Phones are now powerful enough that you can stream that signal by Bluetooth directly to the phone, and in this case do activity recognition and calorie counts on the phone. This was the project from one of our students who has now started a company actually to commercialize this.

Some of the experiments that we did there made use of this technology. One was a context where energy management system ñ all of those little boxes in red are devices using power. There’s one person sitting on the couch. Over a two-month period we did activity recognition every 8 seconds and then we built an algorithm that would ruthlessly kill power to anything that wasn’t in active use, and so here you can see everything being killed, except the lights in the living room and the TV and the refrigerator. We came up with a 42% reduction over that 2-month period. Of course, it killed the clock radio when you left the bedroom, but we think we could probably get to half of that with a fairly reasonable system and I don’t know any more effective way to reduce energy consumption by that dramatic amount.

An example of a system that then we prototyped and built and deployed in 10 homes. Very simple ñ it was a location [inaudible 30:44] thermostat. You carry a mobile phone and it knows where you are in the city. It calculates travel time to home and the farther away you go, the lower the set point of your thermostat. It’s just like a seesaw. The challenge was to develop a system that required no programming. Your home was always comfortable and saved energy and we found that, given that only 30% of people with programmable thermostats use it in program mode, we could do better with this system on average.

We’re very interested in persuasion. So how do you present people with information about the potential savings with respect to air conditioning? We would, from time to time, in this experiment present people with an interface, something like this. You could check that little middle box and get 10% savings by bumping the thermostat one degree higher in the summer and then we can experiment with different ways of presenting information. So what’s more effective, $3 today and $1,000 this year, $6,000 in 5 years and $22,000 at age 65? I mean, it’s an interesting question because there’s no obvious way to do this because people discount the future value of things. So it’s an ongoing research experiment in our lab.

We also are looking at robotic facades, the equivalent of the robot wheel. We think that all the mechanical should move out to the perimeter into these prefabricated components that bolt down to the buildings that are mass customized. In this case we incorporated a little articulating mirror that could throw sunlight anywhere into the space. A lot of the energy consumption in the building is heating load from unshaded glass. Where you could shade the glass, then you don’t have sunlight. In this case, you use the mobile phone interface to put sunlight anywhere you want in the space. In this case she is putting it above her head to put soft interreflected sunlight on the counter when she’s working, and maps that to an activity.

In this case, Ronan, who developed this system, has mapped sunlight to cooking. Here are the little sensors that we’ve deployed around. These are collecting the data about the activity. He put one on the microwave and he trained it to recognize microwave cooking activity. For some reason, he chose to washout the display of the microwave with sunlight, but that’s his choice because it’s his personalized sunlight. We’ve done the calculations and we think that at scale this actually could save quite a bit of energy and then throw a very beautiful light deep into a space.

The final project I’ll show ñ if you could turn the sound up, if it plays. I hope it plays. Maybe not. If you could hear this ñ if you could see the video. This was a natural language learning tool. They use the same sensor deployment, but in this case you pick up an object and the home gives you the Spanish equivalent of that or a sentence that uses that object. It’s using technology that we would deploy for health reasons or energy reasons. We’ve done medication reminder systems, studies.

ëI sit in the chair. Table.’ ëI brush my teeth.’ ëMe siento en el sof·. Yo uso el control remoto. El libro es hecho de papel. La computadora es hecho de electrÛnicas. Good night.’ That’s a natural language learning tool. The point is here that we don’t believe in single-purpose technology. I mean, this infrastructure should be used for many things and it should be used to enrich people’s lives.

So cities of the future, again, start with a focus on people, do good design, use technology when it’s useful. Thank you.

Leave a Reply

Meeting of the Minds is made possible by the generous support of these organizations: