Algorithm speeds up bicycle delivery

Page: 8

audio editing Mariana Garcia

Image Editing Alex Calixto Paulo Cavalheri Luis Paulo Silva

Photography Antonio Scarpinetti Ivson Miranda/Señoritas Courier

Seven women ride bicycles on a wide avenue in an urban area. They all wear similar black and green sportswear. They stand side by side, smiling and interacting with one another. In the background are tall buildings and, even further back, the Altino Arantes Building (also known as Banespa), an iconic São Paulo landmark. — **Technology distributes work equitably, presenting personalized alternatives that can be the shortest**

Algorithm speeds up delivery by Bike

Research conducted by a computer engineer received a human rights award from Unicamp and the Vladimir Herzog Institute

When he entered his master's program, computer engineer Gustavo Gonçalves sought to use cutting-edge technology to develop a tool based on solidarity values—not capitalist logic. As a result, he created a program to generate routes with the bicycle delivery cooperative Senõritas Courier, formed by cisgender and transgender women and based in São Paulo. His algorithm reduced the time spent on this task—which previously took one to two days to complete, but now takes just a few seconds. The result of a collaborative effort involving the Technology Center of the Homeless Workers' Movement (MTST), his master's research won the 5th Academic Recognition Award in Human Rights from Unicamp—Vladimir Herzog Institute, in the Exact Sciences, Engineering, and Technology category.

The study, conducted at Unicamp's School of Electrical and Computer Engineering (Feec), began to take shape after Gonçalves had a defining experience upon entering the job market, one that led him to rethink his career and, especially, the use of science and technology in the workplace. At an automation company, the researcher attended a meeting with his employers and was congratulated for being part of the team responsible for developing a tool that enabled employee layoffs. "I was deeply moved by what society expected of my work as a code developer," he recalls.

The impact sparked Gonçalves' interest in deepening his knowledge of automation, a period during which he became acquainted with the phenomenon of the platformization of work and the concept of solidarity technoscience, which describes the use of science and technology in creating a solidarity economy based on values such as sustainability and social justice, not solely on the laws of the market and capital. The researcher also became acquainted with the concept of platform cooperativism, according to which the platformization of work should be based on the values of collective labor.

A young man with fair skin, curly hair, and a full beard wears round-rimmed glasses and a plain brown t-shirt. He sits with an attentive expression and his hands raised, as if explaining or emphasizing something during a conversation. The background is a simple beige wall. — **Developer Gustavo Gonçalves, who created the algorithm: experience in the job market influenced his interest in the solidarity economy**

Along the way, Gonçalves met Professor Romis Attux of Feec (Feec) and enrolled in the Master's program in Electrical Engineering. His research, conducted at the Signal Processing for Communications Laboratory (DSPCom), was supervised by Attux himself and Cristiano Cruz, a professor at the Mauá Institute of Technology. To support his work, he studied the emerging concept of worker-owned platforms, created by Brazilian Rafael Grohmann (a professor at the University of Toronto, Canada), and popular engineering, a movement that emerged in Brazil and gained popularity in the 2000s.

The invitation to participate in the development of a delivery platform for the Senõritas Courier cooperative came from Aline Os, the collective's president. Gonçalves was tasked with developing an algorithm to generate delivery cycle routes, taking into account the time spent on the journeys, a more equitable distribution of work among everyone, with maximum weight, volume, and distance limits for each cyclist, and respecting the particularities of their bodies and different moods. The now master's in electrical engineering understood that creating routes for the cooperative's members wasn't a simple bureaucratic task, but a moment to understand how people were feeling. Therefore, it was also a work of care. "There are bodies that menstruate, older bodies, bodies of various types and shapes—something that computing typically abstracts, so that the entire body becomes a standard body, considered perfect," observes Gonçalves. In the researcher's opinion, it's difficult to bring this care work into the computer. “When we understood this, we realized that our solution had to help them in this process, as it could never completely replace them.”

A widely used methodology for finding solutions to vehicle routing problems, the genetic algorithm—so named because its operation is inspired by the evolutionary process and natural selection—was chosen by Gonçalves. According to Attux, the tool was already known to work well in solving combinatorial problems involving route establishment. "The general flow of the genetic algorithm's operation seemed educational enough to allow us to explain more or less what the solution methodology would be like and even to be able to work on it with the cooperative members, without having to go into the mathematical details of how the problem is formulated. We thought this could be beneficial because it would allow for an empowering perspective, of wanting to discuss how the algorithm works, so that these people can take ownership of it," says the Feec professor.

A middle-aged man with a beard and dark hair wears glasses and a green t-shirt. He is seated and smiling slightly, with his hands raised to table height, in a gesture that suggests he is explaining something or having a relaxed conversation. The background is light and neutral. — **Romis Attux, research supervisor: the methodology used allows users to take ownership of the algorithm**

The algorithm created by Gonçalves helps ladies to distribute the work among themselves more equitably, presenting an alternative that can also be the shortest. “In the capitalist view, if I can charge the client only for the period that is important to them, that is what bikers will receive. Señoritas, on the other hand, considers how much they're pedaling," says the computer engineer. "We understood, during this process, that we're living in a time when artificial intelligence wants to enter everywhere and replace all human action. If the process can be automated, it seems it should be automated, and our idea was not to completely automate the care work performed by the cooperative members, but rather to introduce a tool that would assist in this work."

The solution involved processing data from previously placed delivery orders, already categorized by Señoritas, and gathering suggestions from the cooperative members to test and fine-tune the algorithm. This allowed the researcher to customize the tool. In addition to each member's name, the data used took into account how much each member could carry, in terms of weight and volume, as well as the type of bicycle and the maximum distances they were willing to travel. According to Attux, the results stand out not only from a technical perspective. "There's a bias at the university when it comes to outreach work, and Gustavo shows that, methodologically, it's possible to do cutting-edge, high-level engineering without having to follow the traditional metrics of capitalism and human exploitation," he states.