We will work with the Montevideo City Hall to improve the management of household waste.

Together with the Institute of Computer Science (INCO) of the Faculty of Engineering (Udelar), we will carry out an initiative to optimize the services of the Montevideo City Hall (IM).

The project, called SIRR (Intelligent Waste Collection System), aims to improve the cleaning and household waste collection management in Montevideo and will be developed over a period of ten months.

It arose from a challenge funded by the National Agency for Innovation and Research (ANII), in which our proposal was selected from more than 20 participants.

The problem

The Department of Environmental Development, through the Cleaning Division of the Montevideo City Hall (IM), is responsible for waste collection in the city. Currently, the container collection system has 117 routes, each with approximately 100 containers, and around 230,000 tons of mixed waste are collected annually.

For the collection of mixed household waste, a theoretical position of the containers in the territory is used (based on historical data), as there is no real-time information on changes in their position, removal for maintenance, among other factors. This leads to the collection routes of the circuits being planned and executed based on assumptions that may be far from the actual situation.

On the other hand, the container lift record is executed manually, so it is necessary to digitize the information in order to process it, make use of it, and make data-driven decisions.

The challenge: an intelligent collection system for Montevideo.

To address these points, the ANII challenge proposed finding a solution that provides real-time information and develops predictive analysis to optimize the management of household solid waste operations in the city. This initiative aims to reduce costs, improve operational efficiency, customer service, and the dissemination of reliable and timely information.

The expected results for the project are:

• To know in real-time the level of compliance of a route.

• To predict the completion of the route (i.e., when the truck is about to finish its collection route) when less than 50% remains.

• Based on the trend of the predictions, allow decision-making to modify the collection circuits in such a way that they improve their performance.

The selected proposal.

Among the proposals from more than 20 institutions and companies, the evaluation committee chose the one submitted by Isbel, considering it "solid, modular, scalable, and sustainable in the long term." Additionally, they highlighted as a strength "the collaborative team between a private sector company and an academic institution."

The response to the challenge focused on installing the system's intelligence in the truck, rather than placing it in the containers (as is traditionally done). This makes the solution scalable and minimizes intervention in the containers. Furthermore, it represents a less risky investment, as containers are often vandalized.

This data is collected through a system of sensors and software, which will be implemented during the project. Then, this information will be processed using a prediction model, which will be developed by INCO.

To carry out the project, we are looking for four technical profiles: Developer Leader, Junior Full Stack Developer, Semi-Senior Full Stack Developer, and Engineering Assistant.

Some repercussions

The team and the development of the SIRR will be led by José Luis Nunes, Product Line Manager of Internet of Things (IoT) & Smart Cities at Isbel. Regarding the opportunity, he commented: "We are proud of the team for being selected among more than 20 companies and entities that participated in the call. We also consider ourselves privileged to contribute to making Montevideo cleaner and more 'intelligent'."

Regarding the project they presented, he stated that it is a "completely innovative proposal in its approach to the solution" and highlighted the participation of the FING "to analyze the state of the art of these solutions at the prediction level."