Container-based Action Routing Tool (CART)

“Access to adequate and equal sanitation and hygiene for all” (SDG 6.2) is universal and foundational to other human rights. Poor sanitation reduces human well-being, limits economic development, and increases the transmission of disease. 

More than 700 million people live in urban areas without basic sanitation amenities, with 80 million resorting to open defecation. The WHO estimates that 1.4 million deaths each year can be attributed to insufficient water, sanitation, and hygiene (WASH) access. Women are disproportionately impacted, and diseases caused by poor WASH access are responsible for nearly 400,000 deaths of children under the age of five. 

The consensus is that radical action needs to be taken to deliver effective sanitation services, especially in informal settlements where traditional sewage systems are often lacking, non-existent, infeasible, or inappropriate. 

One promising intervention is Container-Based Sanitation (CBS), where human waste is safely accumulated in sealed containers before being transferred offsite for treatment. CBS services meet the WHO criteria for safely managed sanitation and case studies have shown it has nearly eliminated the reported use of open defecation in some communities, contributing to public health objectives.

Notwithstanding its evident merits, CBS services are currently not cost recovering, and the key to CBS providing the full sanitation value chain, including limiting the spread of fecal contamination within household and community environments, is keeping costs low and achieving economies of scale. A significant barrier for CBS providers, which are typically nonprofit and/or small social enterprises in scaling up, is managing waste collection logistics. 

Manual scheduling systems result in expensive inefficiencies and leave the hardest-to-reach households underserved. CBS providers estimate that manual route planning processes lead them to make at least three times as many trips as needed which, in fuel costs, adds up day to day. Growing its customer base is essential to a CBS provider’s existence, but doing so results in frequent changes in routes and service windows -- which impacts hundreds of customers - causing significant dissatisfaction, and increasing the potential for customer loss. 

Finally, with ever-growing kilograms of waste to be picked, routes need to be planned with vehicle capacities, capabilities, and terrain all taken into consideration which is too complex to manage manually. 

There are multiple benefits to both business and consumer in a technical solution that eases the burden on CBS providers by helping them plan their waste collection logistics. Such a solution would lower costs and increase operational efficiencies for the CBS provider, while also helping expand access to sanitation and making them safer for the most vulnerable customers.

To help CBS providers optimize their collection services, DataKind developed the Container-based Action Routing Tool (CART).

CART is a service equipped with planning algorithms that automatically distribute assignments across the CBS transportation fleet, creating optimal routes. Leveraging software from OpenStreetMap and Google OR-tools, CART answers the ‘capacitated vehicle problem’ and designs a route optimization tool for the local geography of the implementation partner. 

Developed through sequential collaborations with three members of the Container-Based Sanitation Alliance (CBSA) - SOIL (Haiti), Sanima (Peru), and Sanergy/Fresh Life (Kenya) - CART considers a multitude of local factors essential for effective planning.

Benefits that CART users have reported to DataKind: 

• Reduced miles driven, leading to lower vehicle wear-and-tear.Significant savings on fuel costs, which typically account for 60% of a fleet’s operating expenses.
• Decreased labor costs, due to reduced overtime work by drivers.
• Time savings for those who previously planned routes manually, which can be channeled to other needs
• Enhanced productivity, allowing for more regular deliveries, pick-ups, and service calls - which improves the service and experience for CBS custome

By optimizing collection routes, CART helps CBS providers become more efficient in several ways: by driving down operational costs and delivering a better service they can exceed customer expectations and grow their business. Finally, expanded sanitation services reduce disease transmission, which reduces future demand on health systems.

Case Study: DataKind first developed CART through an initial partnership with SOIL (a CBS provider) in Haiti. Their goal was to complete their journeys with greater efficiency, so savings from operational costs can be redirected into providing a better service to more households. A year after deployment, where CART was used across seven of their nine areas in northern Haiti, SOIL reported a 10% saving in route distances. This reduction in route distances enabled SOIL to serve more families. Further, SOIL also used CART to onboard new drivers, and re-assign existing drivers to new areas, with ease - helping them scale their operations.

CART is an upstream solution for expanding container-based sanitation services, especially in informal urban settlements. 

Deploying CART serves several populations:

1. Residents of informal urban settlements (including slums) where traditional sewage is not feasible. The number of people in informal urban settlements lacking access to safe sanitation varies widely depending on the region and specific circumstances. However, it's estimated that globally, hundreds of millions of people living in informal urban settlements do not have access to safe sanitation facilities. CBS is a viable low-cost solution for providing the safe collection, transport, and treatment of excrement. CART helps keep CBS costs low, which allows the operating model to scale to reach more residents. CBS meets the WHO criteria for safely managed sanitation, so increasing sanitation in this way will support the lowering of disease burden attributable to inadequate WASH access, and prevent more deaths. 

2. CBS providers. The majority of social impact organizations (SIOs) working on the frontlines of the world’s most urgent challenges believe that cutting-edge technologies are beyond their reach, as they do not have the technical skills, the available resources, or the internal capacity to incorporate a more data-driven approach in their work. DataKind closes this technical gap, serving the social sector, by partnering with SIOs and developing technical solutions that advance their theory of change and amplify their impact. CART is an open-source solution that we are making freely available to CBS providers, so they can operate more efficiently and extend sanitation services to more people.

3. Sanitation workers. Sanitation workers deliver the essential sanitation services that keep streets, rivers, and beaches free from fecal waste. However, typically sanitation workers are marginalized and face considerable challenges; including exposure to health hazards, financial insecurity, social discrimination, and even exclusion. This is especially true in low and middle-income countries. In 2023 in an interview with the Exec Director of CBSA, DataKind learned that CART has helped to enhance the standing of sanitation workers and their profession in the communities they serve by empowering them to better serve their designated areas, better reach customers, provide an efficient service and better know their areas.

Health systems in LMICs. Healthcare providers in LMICs frequently contend with financial constraints and scarcities in medical resources, compounding the strain on their healthcare systems. This burden is further amplified by the elevated prevalence of communicable diseases within these regions. The implementation of CBS initiatives enables providers to broaden their sanitation services, thereby mitigating their clientele's vulnerability to transmissible diseases. Consequently, this proactive measure serves to alleviate the future burden on local healthcare systems.

DataKind is the best organization to deliver this solution as we developed and then expanded the original CART solution. Further, we have the technical talent and leadership to grow the solution, and the appropriate community-embedded network to inform and support the development as well as implement the solution at scale.

DataKind is a global nonprofit that harnesses the power of data science in the service of humanity. We are issue agnostic, as we believe that cutting-edge technologies (including data science and AI) can play an outsized role across the social sector, and should play a key role in helping humanity achieve the SDGs. However, DataKind recognizes that our expertise is in identifying how to use ‘Data and AI for Social Good’ and developing data-powered tools and resources to achieve that, and NOT in direct service delivery. It is for this reason that DataKind operates through direct collaboration, co-creation, and communication with social impact organizations (SIOs) on every project. 

Collaborating with a frontline service delivery partner, typically embedded in the community they are serving, helps ensure both the adoption and utility of our solution. This frontline partner acts as the technical project’s co-creator, primary user, and implementation partner - and working so closely with them from scoping to handoff phases ensures that we are truly ‘designing with, not for’.

With CART, DataKind has partnered with three individual implementation partners (SOIL, Sanima, and Sanergy/Fresh Life) as well as one global network partner (the Container-Based Sanitation Alliance) to ensure we’re creating a solution that is centered on the needs and experiences of those CBS providers (and particularly the drivers) operating on the frontlines and serving the communities directly. This existing network - who we have been working with since 2017 - will continue to be an invaluable resource in the future development and further scaling of CART.

The programmatic work on this technical project would be overseen by Mitali Ayyangar, DataKind’s Director of Product & Programs. Before joining DataKind Mitali was a program manager for community-based health programs with Médecins Sans Frontière, and Slum Dwellers International. A longtime part of our CART team is Sébastien Ouellet, a data scientist with GIS (Geographic Information Systems) expertise who has been working with DataKind since 2017.

We believe that CART is at the Growth stage, as over multiple phases of development, testing, deployment, refinement, and expansion, DataKind has developed CART to be a functioning tool in three social impact organizations, in different geographies. The primary barrier to deploying within all organizations in the CBSA is a funding challenge - not a technical challenge.

Our work has developed CART to a point where users can:

• Create a route through hundreds, or thousands, of locations in minutesChoose to minimize either time or distance across trips
• Set specific time windows per area, or per location, to be respected
• Configure routes for multiple different vehicles with different speeds, allowable roads (e.g. steepness, surface types, etc.) and capacities via custom profiles
• Produce detailed maps of the trips in html files
• Exports GPX tracks to be used in turn-by-turn navigation tools (e.g. OSM Automated Navigation Directions at https://osmand.net/)

SOIL Haiti - our first SIO partner, who co-created the initial version of CART - has many hundreds of customers to visit every week and has been using the tool for 4 years regularly.

DataKind is seeking philanthropic funding to continue to expand CART as a service so it can be adopted by all organizations that wish to have it incorporated into their operations. We’re hoping that MIT Solve will assist DataKind with introductions to both philanthropic funders that are a strategic fit for CART, as well as connecting with any relevant contacts that can help us think about expanding our funding aperture beyond philanthropic funding and toward subscription models, software collaboratives, or other approaches.

CART is a truly innovative solution. 

Firstly, its ease makes it very accessible to all users with a simple yet visually intuitive workflow, while simultaneously offering advanced automation capabilities tailored for technical users, ensuring seamless achievement of their IT infrastructure objectives. Moreover, it performs well at scale because it leverages open-source libraries and meticulously crafted algorithms to deliver unparalleled efficiency even in the face of immense data loads. This prowess is complemented by its flexibility, both in pre- and post-processing stages, due to the seamless integration of domain expertise via our tool, which adheres rigorously to the standards set forth by OpenStreetMaps and Open Source Routing Machine. 

Furthermore, our solution is imbued with heuristics meticulously crafted through collaborative efforts with organizations deeply embedded in the community, addressing specific pain points felt by users. For instance, organizations often grapple with the dual challenge of meeting the expectations of both their workforce and clientele regarding service delivery. In response, we have engineered our solution to mitigate such concerns, ensuring optimal efficiency while minimizing potentially disruptive elements, such as the unnecessary back and forth of collection vehicles, which could undermine customer satisfaction. 

Through a combination of user-centric design, cutting-edge technology, and a commitment to addressing real-world challenges, our innovative solution emerges as a beacon of excellence, poised to redefine the landscape of IT infrastructure management.

We have developed, tested and scaled this solution with three partners in three different geographies. We are now linked to a global network (the CBSA) and want to work with them to make this a truly scaled solution

CART aims to help CBS provides deliver provide an efficient service, manage logistics more efficiently and ultimately scale its services. 

To measure the impact of our solution, DataKind will work with leaders at CBS service delivery providers to collect baseline measurements and generate evidence about whether our solution has resulted in efficiency gains. Our implementation partners are committed to co-creating impact metrics such that we are jointly involved in ensuring the investment in a route optimization tool is supporting improvements in health outcomes as well as organizational efficiencies.

Our methodology to evaluate impact will include standard analytic processes that use data collected throughout the project cycle. Qualitative and quantitative methodologies will be leveraged and the scientific rigor of the analysis will vary based on availability and quality of outcome data, which we will have access to through our partners. This outcome data we anticipate using includes; route distances, average route time per vehicle operator, amount of time vehicle route operators spend to complete their service schedule, timeliness, internal operational costs, employee retention rates and customer satisfaction data (n.b. this is a non-exhaustive list).

CART uses:

• Combinatorial optimization: Two open source libraries, or-tools and OSRM, to allow us to define flexible routing with time windows and capacity problems and process road networks efficiently.

• Crowdsourced geospatial data: OpenStreetMaps allows any user to provide updates to the road network so CART can offer solutions that match what’s happening on the ground. This is especially important in informal settlements where local conditions may change rapidly. We offer instructions to ensure our users can provide relevant updates regarding new roads, surfaces that should be avoided, traffic restrictions, etc. This also encourages more organizations to build up the OpenStreetMaps database, distributing their local knowledge to a global community.

• Clustering and ordering heuristics tuned to user preferences: homegrown geospatial algorithms were needed to ensure the tool could scale up and provide intelligible solutions to the organizations.

• Python ecosystem: we’ve built integration points and a containerized server-client model to allow multiple modes of use beyond a command line interface.

5 - 4 FT members of DataKind staff, and 1 pro bono expert technologist

7 years (multiple phases)

DataKind is a women-led, inclusive and equitable, global nonprofit, powered by a global community of technologists. We invest considerable time and effort to ensure our technical solutions are co-created and owned by the community directly impacted.