# Science of Teams: How MIT Media Lab Builds Cities Using Lego and Augmented Reality

https://www.wired.com/video/2017/01/science-of-teams-mit-media-lab/

The MIT Media Lab is using innovation to boil efficient teamwork down to a science. With an enhanced ability to communicate across teams, MIT is creating a workplace that shares ideas in unprecedented ways. The Changing Places group at MIT tackles large challenges like fighting pollution and urban modeling; the latter of which is being solved by using a combination of lego bricks and augmented reality.

Published on Wired, 01.23.17

# Computational Ants: Agent Based Visualization Technique with CDR OD MATRIX

So you want to make a model of a system, but you don't think you have all of the data you need? This is actually a very common situation in modeling. Before going on a quest to search for more data, I often remind researchers that a good model might not need as much data as they think. Personally, I think elegant math models operate with minimal data. The less data you need, the easier it is to deploy, calibrate, and scale the use of your model in the Real World, after all.

In this post we will review a data-light technique for modeling flows of people using a simple and common data set called an origin-destination matrix (OD matrix, for short).

Video: Nina Lutz

### WHAT IS AN ORIGIN-DESTINATION MATRIX?

Trip behavior data is often collected as an OD matrix. An OD matrix reflects aggregate flows between a number of discrete locations over a period of time (Fig. 1).

Figure 1. A conceptual diagram of an origin-destination matrix. Source: Dr. Jean-Paul Rodrigue, Dept. of Global Studies & Geography , Hofstra University, New York, USA.

Unfortunately, an OD matrix does not capture routing between locations. It's also challenging to visualize the weight and directionality of flows for larger matrices. The following modeling techniques, however, can help us recover the missing information.

### Methodology

For our exercise, we'll use a 2D environment created in Processing. An OD matrix is randomly generated (Fig. 2 and 3). In a simple visualization, trip counts are represented by lines connecting points. Larger counts are represented by thicker lines. Color in Fig. 3 is arbitrary but unique to each OD pair.

Figure 2. Randomly generated origin-destination pairs. Image: Ira Winder

Figure 3. Randomly generated OD matrix. Image: Ira Winder

That's great, but now we want to tease out some additional information. In particular, we want a better understanding of directionality and routing within our OD matrix. First, we'll tackle directionality using agent-based modeling techniques. Solid lines are replaced with continuous flows of individual agents (Fig 4).

By animating the network, we have a bit better understanding of flow, particularly directionality. The total amount of agents flowing between two points at any given time is proportional to the flow defined in the OD matrix. Keep in mind, however, that time is an abstract dimension in this visualization. Motion and speed are in fact arbitrary constructs that only serve to make our OD matrix more human-readable. The quantity of agents on the screen, however, is not arbitrary. Rather, quantity of agents is the key parameter by which we communicate information from our OD matrix.

Figure 4. Agent-based modeling technique that communicates relative flow and directionality of a random OD matrix. Image: Ira Winder

Figure 5. Pathfinding algorithm applied to approximate routing. Image: Ira Winder

Now that we've tackled directionality, we'll try to tackle the issue of routing. You've probably noticed that OD matrices tell us nothing of how one travels from point A to point B. While visualizing the routes as straight-shot lines is easy, it is hardly realistic (except for the birds, perhaps). A pathfinding algorithm, such as Dijkstra's, can be helpful for solving this. We've randomly generated a "road" network for our model, lightly colored in gray. The agents are now programmed to take the shortest route along the network provided (Fig. 5).

Not only does our visualization show a more realistic representation of routing, it starts to reveal emergent "corridors" where traffic, and perhaps congestion, might be prevalent. However, any observations at this point are purely qualitative.

### Poly-Voxels

By the way, you might have noticed that our example uses a poly-voxel network (Fig. 6a and 6b). This is mostly an arbitrary preference. It might make sense, however, given some of our other projects.

Figure 6a. Voxel-based data abstraction. Image: Ira Winder

Figure 6b. A poly-voxel network with shortest-paths between origins and destinations highlighted. Image: Ira Winder

### Network Design

Despite all my talk of minimizing the need for data, you might point out that even a routing algorithm requires data for a road network. Touché. In the example above, we've randomly generated data for a network, but you might want to model a specific network yourself. We've created a couple tools to help with that. One tool let's you manually "sketch" a valid network by drawing obstacles (Fig. 7). We imagine this technique might be useful if you lack vector data but can get ahold of a raster/satellite image.

Figure 7. A custom network is created by subtracting polygons from the field. Image: Ira Winder

Figure 8a. Agent-based visualization of pedestrian activity between amenities. Image: Ira Winder

In most cases, though, road vector data is available from various sources such as Open Street Maps. We made a utility that converts such road vector data into our poly-voxel format (Fig. 8b). There's no particular reason you couldn't use vectors directly from OSM, but we prefer the sketchy-ness of the poly-voxels to emphasize the roughness of our approximations. We tested the OSM network by modeling pedestrian movement between amenities (Fig. 8a).

Figure 8b.  GIS Vector data (left) translated into poly-voxel data (right).  The translation differentiates between different path types such as roads, sidewalks, and pedestrian bridges. The purple network is an elevated walkway.

### IMPLEMENTATION

Our primary implementation for this model was for visualization of call detail records (CDR) for the City of Andorra la Vella (Fig. 9a). CDRs can be converted into a type of OD matrix where origins and destinations are represented by cell phone towers. The transition from one tower to the next is captured when a user's cell phone signal switches from one tower to another. This transition is called a handoff and can be the source of a very rich OD matrix.

Figure 9a: Andorra CDR OD Matrix from Cirque du Solei dataset, representing 1 hour of calls handed off between cell towers. Image: Nina Lutz

One issue with our visualization thus far is the use of cell phone towers as our literal origins and destinations. In reality, people do not travel to and from cell phone towers but rather to the amenities and homes within their coverage. Using a Voronoi technique, we can determine roughly which Andorran amenities are associated with each cell tower (Figure 9b). This is, of course, a very rough approximation.

Figure 9b. Voronoi method used to allocate cell phone tower data to surrounding amenities. Image: Nina Lutz

The result is a more nuanced visualization with trips that begin and end at valid amenity locations (Fig. 9c).

Figure 9c. Agent-based visualization of Spanish-speaking (orange) and French-speaking (blue) tourists of Andorra la Vella using CDR data, Amenities, and road network data. Image: Nina Lutz

The methodology was also adapted for a tangible interactive model of pedestrian activity of a Singaporean neighborhood (Figure 10). Users can adjust an OD matrix by configuring pieces that represent amenities.

Figure 10. A bus stop, represented by two red circles, is not within walking distance of nearby amenities (left). Amenities placed within walking distance of the bus stop increase simulated pedestrian activity in the areas (right). Image: Ira Winder

Contributors: Ira WinderNina Lutz
Special Thanks: Yan Leng, Naichun Chen, Arnaud Grignard, Luis AlonsoKent Larson

Editor's note: The term "Agent Based Visualization" appears in other literature including, but not limited to:

# Lego Logistics: Tangible Interactive Matrix Meets Last Mile Logistics Simulation

MIT Media Lab Changing Places Group and MIT Center for Transportation and Logistics are developing a decision support tool for calculating delivery service areas. Logistics experts can use the platform to present parametric models of logistics in a real-time, changeable environment. Researchers expect the tool to improve collaboration and consensus when optimizing distribution networks for last mile logistics.

Video by Nina Lutz.

The tool uses the tangible interactive matrix (TIM) developed at MIT Changing Places Group. TIM uses an array of optically tagged Lego objects, computer vision, and 3D projection mapping.

Users operate the tool by manipulating tangible objects that represent distribution centers (Fig. 1). All together, the objects represent a distribution network. Meanwhile, algorithms provide real-time performance evaluation of the users’ configuration. Key performance metrics in a demonstration include average delivery cost and customer demand saturation.

Figure 1. Two distribution centers are placed upon the tangible interactive matrix (TIM). Photo by James Li.

The use of both tangible bricks and geospatial models led us to adopt a voxel-based method for data abstraction.  (Note: a voxel is a multi-dimensional pixel).  The result is a mathematical model uniquely structured to be compatible with TIM (Fig. 2).

Figure 2. A typical GIS polygon construct (left) is translated into a TIM-compatible voxel and Lego construct (right).

GIS data such as US Census parcels are processed and cleaned to be compatible with the system at three scales: 2km, 1km, and 500m per pixel.  In this scenario, we use population as a proxy for demand (Fig. 3).

Figure 3. Voxel-ized data can represent different areas at different scales.

Average delivery cost is a function of both distance traveled from distribution centers and the density of deliveries made at the “last mile”. Average delivery cost “C” is proportional to customer’s distance from a distribution center “D” divided by density of customers at last mile,  “ρ” (source: MIT Center for Transportation and Logistics).

C α D / ρ

The “last mile” refers to the short but most difficult last leg of a journey, such as a walk from a subway station to home. In the case of delivery logistics, the last mile can refer to the difficulty of handing off packages to customers at home or finding short-term parking. Cost is reduced when many drop-offs can occur within a small area.

Figure 5. Service area solutions for 3 different placements of a single distribution center.

Customer demand is saturated when a distribution center has capacity to serve a given area.  Service areas are automatically allocated in a global manner such that average cost is minimized (Fig. 5). The result is often a non-intuitive pattern of service areas (Fig. 6).

Figure 6. Complex solution with five distribution centers of various capacities.  Green denotes cheaper areas to serve, while red areas are more expensive.

### Collaborators

MIT Media Lab
Ira Winder

MIT Center for Transportation and Logistics
Matthias Winkenbach
Daniel Merchan

### Special Thanks

Edgar Blanco
Brandon Martin-Anderson
Mike Winder
Nina Lutz
James Li

# Update: CP alum won the inaugural MIT 'Food and Agribusiness Innovation' Prize & Harvard 'Seed for Change' Competition

Update: CP Alum Francesco Wiedemann and his team, won the inaugural MIT Food and Agribusiness Innovation Prize last week as well as the Harvard Seed for Change Competition this week. Wiedemann added that "In both contests it was an honor and a privilege to share the stage with other talented teams with a passion for food, agriculture, and sustainability. We’d like to thank the contest teams, judges, sponsors, and mentors for all of their support. These prizes will really accelerate our ability to get things implemented in India!"

Francesco Wiedemann, who has been a visiting researcher of the Changing Places group for the last 6 months, now got into the Finals of the MIT \$100K Accelerate Competition with 'Gomango'.

Together with his co-founder, Naren Tallapragada, he recognized a pressing need in many developing parts of the world. Refrigerated transport puts fresh food on the table, as long as you can afford it. The developing world can't, because refrigerated frights and trucks are too expensive to own and operate. But what if it would be possible to convert normal trucks into refrigerated ones?

Project's logo By Francesco Wiedemann

The startup 'Gomango' is aiming to develop low-cost, low-power, smart refrigerated boxes that will turn Indian trucking companies into high-value cold chain providers. This solution will prevent food spoilage and expand food choices, driving growth for the startup and for India alike.

Come and support team gomango this Wednesday on February 10th @7pm in 10-250.

# Urban Modelling for Refugees in Hamburg

### How to accommodate 80,000 Refugees?

The current refugees crisis in the Middle-East imposes unprecedented challenges to cities all across Europe. The immediate need for massive amounts of housing, amenities, jobs and services, forces cities to rapid their planning and decision-making process. These cities are facing a twofold planning challenge which resembles the post-war era: from one hand, the emergency of the situation requires fast and easily-available solutions. But form the long-term perspective, a comprehensive planning proccess is necessary for successful integration of these newcomers.

The city of Hamburg, Germany confronts similar questions in the past year. The city is projected to accept 80,000 refugees by the end of the year; Currently, a uniform distribution of asylum seekers poses major challenges to the city’s urban domain. How can a decentralized accommodation of refugees lead to successful integration? That is the main question occupying the CityScienceLab at Hafencity University, a collaboration with the MIT Media Lab Changing Places Group.

Hafencity University

In the coming few months, HCU CityScienceLab will confront these issues through the construction and deployment of tangible-digital city models. These models will allow different decision makers to come together and discuss long-term proposals for thousands of refugees city of Hamburg.  The city has already provided  39,000 places for asylum seekers; The purpose of these tools is to help finding suitable areas for another 40,000 accommodation. From the research point of view, the Lab's goal is to promote an objective and substantive data-driven discussion on land use in the city.

### Urban Modelling Platform

The city models consist of two tables: On one table, a model of the entire city of Hamburg presents previous asylum seekers accommodations. This model depicts existing and planned facilities for initial reception and the follow-up accommodation and planning frameworks  (for example, the plan designation of an area). Additionally, all district  and neighborhood  boundaries and important landmarks such as the Elbe, Alster and Bille, are featured as well.

but the overall city scale allows only a rough analysis of the area. Therefore, another model was built to zoom in on the scale of the neighborhood, showing streets, existing buildings and parcels. HCU researchers developed cartographic data base, where potential public open spaces, as well as hard and soft criterias for each plot are presented. This multilayered information can assist in finding other locations for asylum seekers accommodations in preliminary assessment stages. Together with various decision makers, each parcel could be analyzed and check for feasibility.

Replacing different 'data blocks' in the model will allow real-time increase or decrease of the numbers of accommodations at individual location. These real-time changes, initiated by decision makers and the participating public will have a direct impact on different parameters in the model, such as the number of accommodated asylum seekers in the entire city or the number of asylum seekers per district or neighborhood.

Cityscope @ HCU

The aim of this work is to promote a citywide dialogue which deals with the question of how to accommodate the present fugitives and those who are expected to arrive in 2016. It aims not to discuss only individual sites, but to promote a discourse in the context of different interests (residential / commercial / conservation) and legal planning requirements. This also allows citizens  to contribute their expertise and discuss the advantages and disadvantages of the given proposals.

An urban simulation laboratory for rapid prototyping, the City Science Lab at HCU are joining forces with the Changing Places Group at the MIT Media Lab for the study of these research questions. Changing Places Group is building tools that support urban planning processes by visualizing location-based data and enabling real time simulations as well as evaluations of urban scenarios. The Cityscope platform developed at MIT consists of several components: the table, on which Lego bricks are positioned as data units; Projectors, augmenting different data on the model and a screen that displays the additional information and an overall view. The open-source nature of the Cityscope project allowed HCU researchers to rapidly integrate this system and to easily modify it to the specific needs of modeling asylum seekers in Hamburg.

MIT Media Lab researchers in HCU student's workshop, Jan. '16

### Public Participation

Starting in mid-April 2016, open discussions and professional moderation by “steg - Stadtentwicklungsgesellschaft”, a private urban development and communication company, will be held at HafenCity University. These will take place in cooperation with refugees coordinating staff and the Senate Chancellery of the City of Hamburg. Events will discuss both the district and the neighborhood level, along the side the implications on the City of Hamburg. Groups of 50 participants are invited to visit HCU and interactively view and discussed the appropriate framework for the city’s challenges.

### CityScienceLab at HCU

Supported by the Free and Hanseatic City of Hamburg (FHH), the CityScienceLab was established in the summer of 2015 at the HafenCity University in Hamburg. The lab is directed by Prof. Dr. Gesa Ziemer in collaboration with MIT as a research unit for the study of cities in the digital era. The CityScienceLab works with stakeholders  from business, politics, civil society and the academy, in order not only to explore changes of cities, but also to constructively assist in their development. Specifically, the lab  studies and scientifically analyzes urban processes in Hamburg. The city models are developed in order to discuss complex issues with professionals and non-professional alike, with the goal of clear representation of significant urban question.

Translated and edited from Hamburg City and HCU CityScienceLab press release, Feb. '16
Further information about City Science Lab: www.hcu-hamburg.de/research/citysciencelab

Cover photo by: picture-alliance/dpa/D. Bockwoldt
Other photos by Ariel Noyman

# CityScope Featured in Presidential Report on Technology and the Future of Cities

The President’s Council of Advisors on Science and Technology (PCAST) recently published "Technology and the Future of Cities Report to the President" that highlighted the CityScope Project developed by Ira Winder.  Excerpt from the Report:

City dashboards that utilize data feeds from open and closed sources have become increasingly common within city governments. Mayors use them as a barometer of urban performance across many dimensions including levels of congestion, pollution, crime, noise, waste, and even pothole repairs. Although some are open source and accessible to anyone with an Internet connection, these tools are typically designed for experts, are often one off solutions, utilize only low-resolution data, and do not fully utilize standards to enable scaling. Cities attempting to regenerate neighborhoods have begun to invest in Innovation Districts, in some cases empowered with special economic zone status. This presents a new opportunity for cities to leverage urban data (e.g., through the City Web) to create a new set of tools that can be applied to Urban Development Districts (UDDs) to maximize their benefit. The CityScope project developed by the City Science Initiative at MIT Media Lab is a data-driven, interactive, tangible, 3D urban observatory and urban decision support system (DSS) designed to engage non-expert stakeholders for city development.

As a three-dimensional urban observatory the CityScope combines physical scale models (made of LEGO bricks) and 3D projections of urban digital data to form a hybrid physical-virtual reality platform that enables multiple stakeholders to engage in urban decision-making. The CityScope has two modes of interaction. The first is passive observation and the second is active, participatory planning. In observation mode, the CityScope visualizes urban data sets, real-time traffic flows, and social media as well as simulated data such as energy consumption or solar access, so that the users can toggle between information layers (see Figure 1). This allows users of the CityScope to identify potential challenges and opportunities when optimizing existing urban systems.

Figure 1: CityScope with satellite data, Twitter user activity, wind flow simulation, and mobility networks (clockwise starting from upper left).

In active mode, the CityScope allows users to physically move elements of the platform (such as buildings or roads) to simulate alternative urban outcomes. For example, if a user moves buildings onto an empty site, then the CityScope will visualize the corresponding increase in the population density and the effects on traffic, energy use, and the demand on city services (see Figure 2).

Figure 2: CityScope Interactive platform.

# Switching hats for disruption

Day 2 of the Disrupting Mobility Summit was focusing on sharing, governing and design. The morning session discussed the role of technology as it empowers individuals and institutions to share and redistribute excess capacity of goods and services. In the session on governing for disruption, the panel comprising representatives from local and national authorities (who repeatedly denoted themselves as “idiots” referring to yesterday’s comment by Nicholas Negroponte on the public opinion of civil servants) discussed the need of changing the norm in governing in order to enable innovation – both on the local as well on the national level. A holistic approach and the collaborative development of ideas were deemed essential for engaging the public and the private sector to achieve disrupting solutions and share different viewpoints. Also in the following session on Designing for Disruption (with two very inspiring talks on robots and designing for the future) the panelists agreed that “switching hats” – working in multi-/inter- or anti-disciplinary teams with each expert focusing on totally new areas – can boost innovation on the power of imagination.

The summit concluded with a closing talk by Zipcar CEO Kaye Ceille, followed by a casual joint lunch with lively discussions and making new friends and contacts. Finally, the participants rushed off to different company visits and campus tours, like to the Media Lab. Great closing for a great summit!

# The Urban Autonomous Delivery Hackathon

From November 6 to 8, 2015, the Changing Places 48-hour Hackathon invited mixed student teams to make up their minds about Future Autonomous Deliveries in Urban Environments.

While in public discourse, the main stream of reflections revolve around future roles of  what today is the passenger car, the very same urban space in which self-driving vehicles will operate also needs to accommodate cargo deliveries. The sub-domains in which participants had been invited to research were as diverse as "Small Packages", "Food&Beverages", "City Services", "Medical Services", "Disaster and Emergency Handling", and finally "Moving People" - the latter being understood as a very special "freight".

By Sunday noon, the hackers came up with an amazingly wide range of solutions - from dense urban cargo pipe networks for small express deliveries to robots crawling through post-quake debris and bringing water relief to victims awaiting their rescue, from route optimization for trash collection trucks to freight cars on subway trains and to resilient networks of emergency building signage. The solutions presented to the jury on early Sunday afternoon reflected a deep understanding for future challenges awaiting urban communities, and included approaches from all fields of advanced science.

# Awards, books and robots

The first day of the Disrupting Mobility Summit ended with a poster session comprising 30 posters covering different topics such as technological innovations as well as social aspects. After that, participants were invited to join a reception with award ceremony for Best Posters and Hackathon winners. The Best Poster Award including two books from MIT authors went to Ronnie Kutadinata and his colleagues from the University of Melbourne for their poster “Shared, Autonomous, Connected and Electric Urban Transport“. The poster presentations of Dewan Karim from the City of Toronto („Innovative Mobility Master Plan: Connecting Multimodal Systems with Smart Technologies“) as well as R. Sheehan et al. from the Federal Highway Administration („Accessible Transportation Technology Research Initiative“, a last minute submission which didn’t make it into the program) received special acknowledgements.

Of course also the winners of the Autonomous Urban Delivery Hackathon have been celebrated during the event. Of the 82 hackers in nine teams, who had been hacking 48 hours right here at the MIT Media Lab the weekend before, 4 winning teams were welcomed on the stage: Team “Capsule” (Michal Cap, Kasia Marczuk, Samitha Samaranayake, Valerio Varricchio) and Team “Dirty Jobs” (Sophia Yang, Angad Randhawa, Hosea Siu, Sesha Sendhil, Christian Umbach) shared the 3rd Prize; the 2nd Prize (awarded by the Hackathon participants) went to Team “Mesh Egress” (Ryan McLaughlin, Jamie Farrell, Siddharth Gupta, Zach Hyman, Xinhui Li). The Grand Prize was well-deserved by Team “Relay” (You Wu, Max Qu, Esya Volchek, David Wang), who developed an innovative approach of combining drones and public transport for urban deliveries. This performance was also rewarded with the Turtlebot2 the team used for simulating drone-bus interaction, which was impressively demonstrated in a video.

The award ceremony ended with the announcement of the Disrupting Mobility Book, which is going to be published mid 2016 in Springer Lecture Notes in Mobility. Authors will be invited from the presenters of the summit based on the quality of their posters and talks. The evening was finally concluded in a relaxed atmosphere with “Relay’s” robot serving drinks for the team.

# Cities are for people… not machines

The Disrupting Mobility Summit kicked off with welcoming remarks by Ryan Chin (MIT Media Lab), Florian Lennert (InnoZ), Philipp Rode (LSE), and Susan Shaheen (U.C. Berkeley), followed by a a series of inspiring talks and a lively panel discussion (moderated by Greg Lindsay, New Cities Foundation) among Dan Doctoroff (Sidewalk Labs), Victor Mendez (U.S. Department of Transportation), Nicholas Negroponte, and Kent Larson (both MIT Media Lab) on the future of mobility and cities in general.

Regarding megatrends like urban growth and technology revolutions, the keynote speakers agreed that the main goal of all efforts being taken should be to achieve high quality of life for all people. Current trends like sharing economy, connected automation of transport, and new housing concepts have the potential to reshape the cities towards more compact and more livable environments.

The (all American) panel also looked into other parts of the world, emphasizing that European traditions like governmental regulations or radical trends like car-free cities could inspire a paradigm shift in the perception of the value of taxes for the society (“normal market forces screw cities”) or the role of the automobile.

About 300 registered attendants from research, industry and municipalities participate in this event. The summit will continue today with sessions on technology and social trends disrupting mobility as well transforming cities.