Stratio was founded in 2015 by Rui Sales and Ricardo Margalho, two tech entrepreneurs, to be the world’s leading Real-time Predictive Fleet Maintenance Platform. The company’s proprietary technology combines large-scale data processing with the latest machine learning techniques to prevent thousands of breakdowns from happening, thus saving millions of people from the hassle of transportation delays. 5 of the 10 largest transportation companies in the world rely on the Stratio Platform to eliminate unplanned downtime and increase operational efficiency. Fleet operators in Europe, North America, Asia-pacific, and Latin America trust Stratio’s technology to fully leverage the data under the hood to safeguard operations, and keep customers happy. Stratio’s technology has enabled transportation for 1.3 billion people so far. In November 2021 Stratio announced a Series A investment of 12 million USD. Currently, Stratio has 80 employees, with more than 50 in R&D areas, and plans to expand to more than 100 employees by the end of 2022.
With the ELISE funding, Stratio will research a new and more efficient way of developing models used to predict upcoming vehicle failures. The project will test the approach on one component of a truck, namely the cooling system. The ambition is to then generalize the concept to other components and enable Stratio to scale up the model development, building a strong competitive advantage in its field.
“Limiting the scope to only one subsystem will increase our chances of getting results within the time frame of the project. Targeting a component that exists in all vehicles and has a relatively high likelihood of failure, such as the cooling system, increases the chances of having plenty of examples to validate the model” says Paulo Bajouco, Stratio’s CFO.
The outline of the ELISE-related project is to develop a self-supervised approach that is also robust. “It has to be self-supervised as there is more data than can be analyzed and annotated by humans. Robustness is required due to the intrinsic nature of the vehicle data previously explained. The models are trained on enormous time series datasets collected onboard vehicles. These machine learning tasks are complex due to the difficulty of finding anomaly patterns in the data. Real-world vehicle data suffers from data corruptions and noise, which has a negative impact on the training of traditional machine learning systems and poses a significant challenge” explains Bajouco. The project also provides the AI community with a good example of how interdisciplinary machine learning techniques can be applied to solve today’s real-world problems. It identifies where existing methods are lacking in order to reach widespread adoption. In turn, this can be addressed by the research community.
Building on results from the ELISE project, Stratio will continue to research and develop improved ways of working. According to Bajouco, “ELISE provided the initial funding that allowed Stratio to explore a radically new approach, which the Stratio R&D team will continue to improve, both in-house and through future EU-funded research projects”. As AI and machine learning are at the core of their product offering, the company intends to continue pursuing research and development and state-of-the-art advancements in this field, leveraging both public funding and private funding. “Within the next 3-4 years we intend to have in production the results of R&D that enable us to develop robust ML models that are not so dependent on the existence of labels,” says Bajouco.
More about the company and their services is available on their website.
More about the ELISE 1st Open call and awarded companies is available here.
