In our previous post on waste management, we explored the causes and effects of unaccounted for solid waste (SW) in cities in low-income countries and briefly introduced our idea of leveraging satellite and drone-based imagery to address this challenge. Since then, we investigated this issue even deeper, by conducting in-person interviews and organizing a virtual focus group with key stakeholders (figure 1).
Over the past few months, we have been collecting valuable insights in the attempt to minimize assumptions, determine the relevance of our initial idea, and of course, to identify the critical assessments in terms of technical feasibility and viability of our proposed solution. In this post, we report our learnings and provide an overview of our next steps.
The Municipal Context
In the first part of our virtual brainstorming discussion, we spoke with two solid waste management representatives from two very distant cities: Chato, Tanzania and Hyderabad, India. While very different areas based on a long list of parameters (size, location, population figures, etc.), the two representatives did both discuss some similar challenges related to prioritizing solid waste management and the lack of sufficient municipal accountability needed to effectively address this challenge.
The Chato township in northwestern Tanzania, an area representing a population of roughly 85,000 people, faces extensive solid waste management challenges (largely domestic, biodegradable waste with some plastic), predominantly due to the poor collection and transportation methods for disposal of solid waste, a reality resulting from location of very limited dumpsites being far from the town center itself. This, combined with fast urban population growth rates, fuels the need to improve sorting methods in order to address the high risk of disease outbreak, especially given the town’s proximity to the significant Lake Victoria water source.
Meanwhile, Hyderabad, a massive urban center in the southern Indian state of Telangana, is home to nearly 9 million inhabitants, including an alarming growing rate of those living in slums. According to the representative, waste generated from households and industries is not properly handled due to the lack of government and local authority accountability. Additionally, waste collection and disposal fees are billed monthly, irrespective of the usage itself/level of sorting required. In such crowded and highly dense areas, solid waste challenges are increasingly difficult to address, as there are limited available dumping sites. Moreover, the high volume of waste poses sorting challenges due to the widespread use of a single bin for all types of waste. Much of the waste itself consists of food and plastics, with the latter serving as a prime market (in addition to newspapers and scraps) for informal collectors looking for items that can be sold. For the waste that does end up in landfills, which are often overfilled and uncontrolled, it makes it even harder to have a clearer picture of what type of waste can be found and perhaps turned into a resource.
These two urban areas are just examples of a phenomenon experienced by countless others. Their common denominator is the lack of sufficient municipal accountability, which often results in waste management operators and municipal leaders viewing efficient solid waste management as a cost, rather than as an opportunity. It follows the common approach of collection and sorting being often left to informal (presumably criminal) groups, which view this as an opportunity for monetization in this uncontrolled field.
However, while Hyderabad is experiencing an increasing amount of solid waste dumped on existing piles of waste, Chato has resorted for the most part, due to limited dumping sites, to burn much of their solid waste in open areas. Additionally, they have made a strategic decision to move their main existing dumping site, which is currently near a residential and school area, to a larger land area well outside of the town center. Given this finding, we believe that our initial focus for a pilot will be set in Hyderabad (or a similar type of large urban setting) as a starting point for the proof of concept stage.
Bird-Eye, A Data-Driven Solution for Waste Management
In our discussion with field and technology experts, we collected valuable insights to refine our award winning solution. In short, Bird-Eye is a data-driven decision making tool for municipalities and waste management operators. By utilizing satellite imagery and camera-equipped drones, different materials in a dumpsite can be detected, as well as their quantity. This data is leveraged for waste management prioritization and shared on the online marketplace, where other municipalities and businesses can bid and purchase specific materials they are in need of (e.g. plastics, biodegradable material, etc.). But how does this actually work?
As shown in figure 2, through cloud-based satellite images, dumpsites are identified, localized in the city (GPS coordinates) and their area (sqm) is defined. A variety of providers, such as DLR and NASA, offer satellite images for affordable prices – depending on the quality of the resolution and the date of the image. According to our research, standard resolution and 1-week old images are very affordable and sufficient to localize and define the area of a dumpsite.
Once the dumpsites are localized, drones equipped with hyperspectral cameras inspect the surface to collect data about its characteristics. Hyperspectral cameras can spot different types of chemicals, as each spectrum has a specific fingerprint – a signature. From this information, we can derive an approximation of the type and quantity of material in the dumpsite. Drones, like the ones provided by DJI, and hyperspectral cameras, like SPECIM, are becoming more and more affordable. For instance, an autonomous professional drone equipped with an advanced hyperspectral camera could cost about $23,000 USD.
While such efforts can help us inspect the surface of a dumpsite, what about what lies under? Solid waste generated from households and industries tends to be constant over time. This implies that a regular inspection of a dumpsite can suggest a good approximation of its contents – A perfect task for machine learning! Furthermore, the repetitive inspection over time/regular intervals can enable predictive analytics. Essentially, we can timely understand if, how and to what extent is waste generation likely to change in a specific neighborhood.
The data collected can be visualized on a dashboard by the decision makers – i.e., waste management representatives (figure 3). Our research shows that the most valuable information for the user is the dumpsite location (coordinates identified via satellite images), its area (m2 also identifiable through satellite images), its volume (m³ identifiable through satellite images or drones), the type of material (spectrum of the chemical identified through hyperspectral camera), its quantity (tons estimated through regular inspection and machine learning) and its market price in the region ($/ton suggested by the system). But why is this information relevant? How can it be leveraged by waste management representatives?
Awareness: Having an overview of the waste (resource) spread across the city helps to detect, even predict, dangerous situations. For instance, toxic waste in sensible areas, such as water sources, and insufficient landfill capacity, are events that can be promptly identified, prioritized and solved.
Optimal Capacity: An approximation of the quantity of the critical types of material present in the city can work as a sort of “distributed stock”, helping decision makers in identifying those materials that are either in excess (e.g. plastics) or insufficient (e,g. biodegradable waste). In this regard, the objective is to maintain an optimal capacity of each material, relying on predictive analytics, internal necessity and market demand.
Marketability: The adoption of Bird-Eye by several municipalities, businesses and informal collectors in the same and in different regions, enables a marketplace that matches offer and demand of different materials. By leveraging a dutch auction model, municipalities have the opportunity to buy, sell or trade materials on the market. Depending on current capacity and waste generation predictions, prices are automatically suggested and updated by the algorithm, adjusting the cost to be higher or lower than the market price. Competitive prices can also increase investments in the collection of the material.
Despite the insightful conversations with waste and technology experts, there are still some unanswered questions. These questions range from technical feasibility to solution desirability and viability. Thus, we strongly believe we need to proceed with an on-site pilot to optimize and fully validate our solution. Specifically, we expect to learn about the following (figure 4).
Desirability: Various waste management representatives in low income countries we have reached out have expressed eagerness to leverage Bird-Eye’s solution for the above-mentioned reasons, but they often have limited experience in SW-based management tools. In this context, although municipalities might be the buyers and beneficiaries, data analysts could be the final user. It is therefore important to define the roles of our key partners, from sales to implementation, from usage to maintenance. The identification of key partners and their roles is also essential to define and prioritize the SW features, in order to select the ones to develop in the short term.
Feasibility: The performance of the hyperspectral camera, in terms of accuracy in detecting and differentiating waste materials, is probably the biggest question mark. To assess this hypothesis, we plan to focus on a specific dumpsite, performing an inspection with a hyperspectral camera first and a manual inspection after. This should give us a good understanding of the delta between detected and actual material on the surface. On the same line, a regular inspection of the surface for at least 2-3 months should suggest whether we can have a good approximation of the volume of the dumpsite, in terms of types of material and their quantity. Last but not least, flying drones are more and more subjected to municipal, county or country regulations. We therefore plan on investigating existing legal and regulatory frameworks, as well as alternative devices (e.g. ground vehicles) in certain municipalities where such a pilot would be more feasible under those limitations.
Viability: Bird-Eye is a social business and, as such, is expected to self-sustain itself in the medium-term. In order to make the solution efficient and generate revenue, we need to consider a smart logistical approach based on an optimal number of drones disposed in strategically located warehouses, accessible to multiple municipalities in the region – in this sense, a “pay-per-use” business model is probably the most consistent. Other fundamental aspects to assess are the best pricing mechanism (e.g. dumpsite area and/or hours of inspection) and the buyer’s willingness to pay.
Bird-Eye’s pilot is expected to last between six and seven months. In this timeframe, we plan to test and validate the solution in two municipalities in parallel, one in India and one likely in Sub-Saharan Africa. Cities such as Hyderabad are ideal candidates in terms of the current waste management situation and challenges, ones that are further exacerbated due to urban population growth. In figure 5, we propose an estimation of the timeframe of each of the above mentioned activities (milestones).
Last but not least, we expect this pilot to identify the KPIs to measure the potential impact of this solution. For instance, we need to verify the average dumpsite volume reduction, the yearly investments in waste management, and the trends of certain types of waste-related disease outbreaks across specific neighbourhoods.
You believe you have resources, expertise or contacts that could be beneficial for running this pilot? Drop us an email! Bird-Eye is a social business that is developing smart solutions for waste management in low income countries. We are always open to sharing our experience and building new partnerships.
This post was co-written with my co-founder and dear colleague Nicola Terrenghi.
This post is based on the professional input and key insights from waste operators actively across emerging markets, hyperspectral imagery and remote sensing experts, and machine learning specialists. I also want to thank the creativity, expertise and commitment of my colleagues from the original founding team: Nicola Terrenghi, Stephen Machua and Tanja Rosenqvist. We worked intensively on“Bird-Eye” during the UNLEASH Innovation Lab, 2017 in Copenhagen where we were awarded the Silver Prize in the “Urban Sustainability” category.