Sukriti - Planning and Creating Healthy and Beautiful Cities

It is an established fact that poorly planned and poorly managed urban transportation leads to significant air pollution, road accidents and increased morbidity and health issues.   

There is a clear case for a better planned, greener and healthier transportation system of cities for healthy cities. 

Our solution proposes to improve the planning as well as management of transportation in cities. This covers both network and mode planning that can be used by city planners and transport operators as well as subsequent management for effective utilization by commuters.

We use multi-layered information comprising commuting dynamics, economics and demographics gathered over a period from real time maps and town planning authorities. This is used to create a node map of city for creating source-destination pairs, commuting routes and selection of modes. This data is then used for conducting pilots and creating a machine learning model for subsequent planning and improvements.

Transportation contributes to more than 25% of air pollution related deaths (~200,000) in the USA [1]), commute to workplace contributes to more than 25% of total road accidents (~0.5 M) in India [2] and sedentary commuting significantly increases the chances of morbidity (more than 32%) [3] and reduces life span.  The ratios would look only worse in developing economies like India.  

In India ~480 Mn of its 1.34 Bn population (~ 36%) [4] resides in urban areas and is expected to more than double to ~875 Mn by 2050 [5].   However, public transport share in work trips is just ~18%. Moreover, the growth of vehicles per million in the last decade was 219% while the infrastructure has increased only by 124% [6].

Hence, a well-planned and managed urban transportation system is bound to play a significant role in reducing air pollution, improving health, safety and quality of life for hundreds of million people in India alone. 

Many green-transport ventures have failed as business ventures for the lack of efficient city transport planning and subsequent management. Our solution not only proposes to provide a starting point in terms of a plan but also a management aid for economic sustenance of the smart transport systems.

Our customers comprise city planners and administrators, transport operators and commuters who are using the transportation and city infrastructure.

The first phase involves working with the city planners to develop a data driven framework and an interactive interface for planning. The second phase would involve working with the commuters and different transport venture owners for developing real time feedback and management framework.

The work on first phase has been taken up with World Resources Institute (WRI) and Kochi (earlier Cochin) Metro Rail Limited (KMRL). Kochi is being developed as a Smart City under a government of India project.

Structured via in-person and audio-video discussions with the stakeholders that have been done over multiple rounds.

The Discussion with KMRL have focused on station catchment areas and last mile connectivity to

         i.  Identify current catchment areas and gaps in connectivity

        ii.  Define criteria for last-mile connectivity

       iii.  Identify pilot points (to check parameters such as terminus, ferry access, socio-economic coverage etc.)

Data has been provided by the Kochi Metro team and has helped us supplement our data which largely consisted of publicly available data sets. 

Our team, Sukriti, was the joint winner in a nation-wide contest organized by WRI , KMRL and TMF (Toyota Mobility Foundation) [7]

The first phase of our solution involves working with the city planners to develop a data driven framework and an interactive interface for planning.   The second phase would involve working with the commuters and transport venture owners for developing real time feedback and management framework, including bundled services for the commuters based on their preferences.

We would discuss the in-progress first phase of the project in detail and the planned second phase briefly.

Our solution is an integration application. The key elements of the application are the following

1. Multi-level layer Plan - We are creating a multi layered node-network map of the station catchment area with existing and proxy data. The layers comprise the following -       

           i. Residential and commercial properties by location

          ii. Home to school/college/education centers

         iii. Hotel to places of interest

         iv. Highway travel routes

2. Using these layers, routes are defined based on commute timings, volume of passengers and demographic profile of customers.  These can be tailored based on places of interest and commuter preferences and demographic profile. 

3. These layers are helping in the following:  Organizing end to end commuting options and Providing greener options to commuters such as walkways, cycles, electric public transport. These inputs are being used for conducting pilots with Kochi Metro team to generate proof points and data for the machine learning model. 

4. Machine Learning Model - We have created a neural network model that uses the data to forecast travel routes and volumes and identify gaps in connectivity. This model would use the factors identified above and the historical route data and the pilots for training. 

 5. Integrated dashboard to forecast trips and bundled service for creation of new services for last mile connect as and where required and strategic partners with a view for a pan Kochi roll out for different commuter groups

The second phase would involve real time monitoring of route performance, providing feedback and access to commuters, service providers and administrators for appropriate action.

Technologies: Deep Learning using Keras and Tensorflow, Django framework for providing APIs, Google Maps for location and real time traffic data, Angular JS for frontend, PostgresSQL for database

City planning has traditionally relied on heuristics and old data from dated surveys, which fail to capture and plan for a city as an evolving entity.   

Our solution aims to bridge the gap using a data driven solution leveraging state of art geographical data mining and machine learning techniques. 

Our innovation lies in selecting the right data sources and using these to model a city. This system is then extended to new use cases using machine learning (ML) techniques that can predict behaviour for relatively new input sets.

We are using historical and near real time data from multiple sources (which essentially form the factors determining the demand for volume and type for transportation) to form a multi-layer model for transportation route planning. This model would be used for the initial planning. The developed model prepares a blueprint for the plan. We then train the ML model for new input sets.

For example, our model considers the following for planning travel routes: residential and commercial properties by location, route for home to school/college/education centers, route for hotel to places of interest, highway travel routes and real time traffic volume data. 

The three key pillars of our technological solution are:

1. Data :- As noted in the previous section, our work is highly data centric. 
   1. We use data from a number of sources to create a blueprint of as is state and train our ML models for future predictions. 
   2. Real time data on transportation usage is sourced from Google maps using APIs and data from web related to people’s preferences is sourced using various data scraping techniques.  This data is merged with quasi-static data provided by town administration authorities.   The entire data preparation stage relies on automated data sourcing, management and preparation for further processing. 
2. Machine Learning :- We use a number ML techniques for extracting information from unstructured and semi-structured data sources, classification, clustering and prediction.
   1. For example, we use NLP for information extraction from blogs and travel sites. The data for this step is custom trained based on our entity definitions and user preference classes.
   2. We use text clustering techniques using a variety of feature definition techniques such as using Glove representation of words.  This helps us identify probable classes which are then used for preparing data for information extraction.
   3. A neural network is being used for predicting new transportation and routing options.
3. Data Visualizaion :- The dashboard and simulation tools are being developed to provide real time feedback for the users. These use various frameworks for data presentation and integration with the back-end system.  

Activities

• Increase adoption of Mass-Transit systems by integrating transport to provide appropriate First-Mile and Last-Mile connectivity

Outputs

Short Term Outcomes

• The consumer moves to the "Next-Best" Green Way of travel (from Car/SUV to A/C Shuttle + Metro Rail, from Auto-rickshaw to Shared Auto-rickshaw) reducing the carbon footprint with similar or better price, convenience or time to travel

Medium Term Outcomes

• Reduced congestion and air pollution

Long Term Outcomes

• "Green-Think" 
  • The city and transport planners incorporate "Green Travel" at the planning stage of any development.  
  • Improved air quality and traffic leads consumers to choose the "Greenest Way" to travel

1. Current number of people -  Kochi Metro Rail has a daily ridership of ~45,000 people, which KMRL wants to increase to ~65,000 people. The first phase which focuses on First Mile and Last Mile access is targeted towards identification of these consumers and offering relevant services (20,000 people).  
2. Number in one year - The passage of the UMTA Act in Kochi will bring all Transport services under the project purview and we expect to serve the entire 2.12 million population of Kochi Metropolitan area. We also plan to be operational in one city/metropolitan area by end of year, i.e. reach to an additional 3-6 million people.
3. Number in five years - In five years we propose to be operational in 10 cities and at least 3 countries (India, China, Indonesia, Brazil, UAE), and hope to impact the lives of 70 million people

Short term goal (within a quarter): We plan to release the beta version of our platform targeted towards the planning authorities within this quarter.  The proof of concept has already been rewarded by the partners and development of features is already underway. We plan to focus on first and last mile access and impact approximately 20,000 commuters

Mid term goal (within a year):  We plan to initiate our work with the feeder service providers post our beta release.  This would help provide integrated services to the commuters and further optimize the end to end solution. We expect the passage of a key regulation in India (UMTA Act) that would bring all transport services under one controlling authority. This would broaden the reach to approximately 2.12 M population of Kochi. 

Long term goal (within five years):  We expect a full roll out of services by the mid of second year. This would cover the planners, feeder service providers as well as the consumers.  We target to move the app to production for consumers by the end of first year. In five years we propose to be operational in 10 cities and at least 3 countries (India, China, Indonesia, Brazil, UAE), and hope to impact the lives of 70 million people

To scale our solution and to increase our brand visibility we foresee the following barriers:

1. Political and Legal Barriers :

Business Barriers: 

1. Marketing: While we are building our relationship with the planning authorities, we also need to reach out to a wide range of feeder service providers and commuters.  The marketing needs for these audience require significant effort and brand building as we move from institutional to retail audience. 
2. Outreach for institutional customer:  We need to establish relationship with a number of institutional stakeholders as they form a key customer segment for our product.  We would need sustained sales and marketing effort to reach out to this customer segment. We would also need expert advice to ensure that we address the requirements of this segment of customers. 

4. Technology: The data sources for information related to our institutional and retail customers vary across cities. New data sources also get added continuously. Our platform needs to be flexible to add new information sources as well as integrate with a large variety of data sources. We also need to ensure that we keep our machine learning component is aligned to fast changing technology landscape to ensure we constantly enhance value for our customers. 

1. Political Barriers : We are collaborating with WRI, India which has wide expertise in Government Interactions and Project Implementations, and will look to expand our team to have local Transportation Experts in every geography we are present in.
2. Business Barriers : Collaboration and building teams for specific roles. For example, we plan to set up an institutional sales team to connect with our business customers. We plan to tap various conventional marketing channels such as the internet, print media and audio-visual channels for our retail customers.   We are also working on establishing sustainable funding for our business to enable these investments and scaling requirements. 
3. General Public : Our model incorporates the customer preferences, by way of income profile, demographics, choice of transport and our Machine Learning Model learns from successes to offer similar choices. At the fist iteration we also plan to offer "The Next-Best" green alternative and look to make a gradual change in commuter sentiment. 
4. Technology:  We have an existing pool of talent comprising members from both technical and management fields.  However, there are very specific technology aspects of our solution that would require specialists in these areas.  For the platform development we plan to hire talent for user interface and user experience development.   We would also need few full stack developers for integrating the platform components for production release. We also plan to partner with universities for expert advice on optimization and planning.  

Sukriti Design Studio has a full-time staff of 6 people with 20+ consultants for 3D-Design, Project Estimation, MEP Services.

Meson Labs, our technology partner has 4 full-time staff dedicated to the project along with 2 Consulting Data Scientists. 

We are an interdisciplinary team who have partnered to deliver a solution which incorporates the best of domain expertise and technology

Sukriti Design Studio is an Architecture and Planning Firm which aims to create responsive design. Specific to this project, we assist in identification of data sets, understanding of urban planning and transport dynamics, and design of visualizations for the end-user.

Meson Labs is a team of engineers, technologists and business consultants, from premier Indian Institutions (IIT and IIM) having technology and consulting experience. We have a focus on technology and end-users, and look to create new ways to seamlessly align the two and bridge the gap between what technology can offer and how users wish to consume it, creating sustainable businesses. 

Sukriti currently has three partners for solution development and initial pilot:    

• WRI Ross Center for Sustainable Cities for technical expertise, research partnership and international operations
• Toyota Mobility Foundation's  expertise in technology, safety, and the environment, partnerships with universities, government, non-profit organizations, research institutions and programs resolving urban transportation problems. 
• Kochi Metro Rail Ltd. (KMRL) for access to key Transport Data and collaboration to  develop innovative design-based approaches to visualize urban transport.

Financial Sustainability

1. (Fee for Service Model) Transport and Urban planning Software / Dashboard for the Central Transport Authority (UMTA)- We aim to offer the dashboard as Software-As-Service or a license based on implementation requirements
2. (Market Linkage Model) - In partnership with the UMTA, the selection of feeder services (type and pricing for optimal capacity) will be offered on fee bases to clients planning to offer said service. These consulting services will also include advertising and partnerships with local and In-house establishments as optimized by the ML algorithms. 
3. (Service Subsidizaton Model) - The end-user interface (planned to be a mobile application) will be offered free to the user to increase uptake of services, and the APIs and development will be subsidized by the software and consulting revenue 

Key barriers to our solution and the benefits we expect from the Solve challenge are :

• Outreach: We are in the pilot phase and scaling of our platform would benefit greatly from peer network and expert advice.  We hope to further these with our selection in this challenge. 

• Branding : Our solution attempts to provide a data rich, immersive experience to various stakeholders - planners, transport agencies and urban commuters.  However, barriers exist in accepting this  solution due to  limited visibility. We expect to leverage the MIT brand to further our inroads with these stakeholders

• Funding: While the current pilot is partly funded by our existing partners, we expect significant cash outgo as we move to the production phase - hosting cost, development and testing cost, brand building and sales cost.  MIT solve prize money would be useful in covering some of these costs

There are two important aspects of our solution that relate to our need for partnership:

1. High data centricity for planning, learning from existing transport solutions and deploying the recommendations for feedback and further improvement
2. Use of state of art optimization and learning algorithms

For these two aspects, we need to:

1. Partner with the urban planning authorities for distribution of our solution and collection of data. This would also help us in monitoring and evaluation of our solution.  We are jointly developing our solution with one such authority. 
2. Partner with feeder service providers (transport operators) for refining our business model. This partnership is also critical for the commercial success of our platform and is potentially the key revenue generating partnership.  The operating model for this partnership would evolve as we start our engagement with this group or stakeholders.  
3. Partner with universities for state of art optimization algorithms for our solution and also source talent

Since our solution is in the pilot phase, some of the partnerships would crystallize as we take our offering to market.  We are working out the modalities of offering our services to commuters who are the last but key link of our solution.   

Our Solution uses AI to predict the travel patterns and Scenario Analysis for changes using the created Data-sets

How you and your team will utilize the prize to advance your solution?

1. Partnerships with Academic Institutions for Optimization of Algorithms being used, identification of key parameters and fine tuning of Algorithms being used. 
2. Exploring Different Use Cases: While the core solution focuses on Home-Office Origin-Destination Pairs (since they constitute >50% of trips), we have found greater success predicting Tourist Trips with Genetic Algorithms. The funds will help us delve deeper and use the best model for the each 

Our solution uses technology to identify slums and the poorest and vulnerable communities in a city and their access to healthcare centers. Unplanned urbanization with the growth of urban slums, lacking reliable piped water or adequate solid waste management, renders large populations in towns and cities at risk.

We believe that a Layered visualization (which integrates Sanitation, Water Access, Income Profiles, to Healthcare Access in an area) will help a city with 

1. Supporting Cities to improve their reporting systems and capture the true burden of the disease
2. Linkage of Parameters like 'Access to water' and 'Sanitation' to disease control and elimination. 
3. Collate evidence for understanding spread of vectors and infection through disease/epidemic instances
4. Develop standardized approaches for readiness and response to major epidemic-prone diseases (quarantines/first response)
5. Visualization driven Community Engagement to encourage behavioural change - The Visualization of decline in Disease incidence when best practices are adopted/not adopted will encourage 

For eg. Abhu Dhabi has 57 Primary Health Care Centers, 13 Hospitals and 12 Specialised Centres managed by SEHA, catering to a diverse population. A simple Layered City map which shows the impact of 'sanitation' to 'diseases' will go a long way towards engaging communities and healthcare workers as initiatives like the Customer Happiness Index has gone a long way towards instilling happiness and positivity among employees and customers.

One of the common concerns when discussing trip planning with Women Travelers, was the estimation of Safety as an important criteria for choice of route, destination. 

We are looking at developing APIs and an App to incorporate crime statistics, lighting conditions for night travel and general safety of routes, to help women travel safely. If we are selected for the Innovation for Women Prize, we would like to explore this avenue further. 

The Sukriti Team is a women-led Team. 

Kochi Metro has been a pioneer in Gender-Responsive Innitiatives. [1] Working with them, and other transport authorities in Kochi, under Phase 2 of the project, we are to enabling "Kochi for all". We are looking at enabling and informing access for women, which would include information on safety, lighting conditions and information on plans for women for different transport channels.

Also as discussed above, while the dashboard has been designed for access to mass transport, the solution is easily extendable to access to healthcare centers and monitoring the delivery of quality healthcare to vulnerable populations through clinics