ITMO and Sberbank: Open Python-library RIDE to speed up logistics calculations for vehicles

2025: Creating the RIDE Open Python Library

Specialists of the Research Center "Strong AI in Industry" ITMO together with the Laboratory of Artificial Intelligence of Sberbank created an open Python-library RIDE to speed up logistic calculations of any scale: from district to international. The development is scalable and will be useful both for the organization of courier delivery and for the transportation of petroleum products. The university announced this on February 3, 2025.

The key quality of RIDE is that it immediately reduces the search area for the optimal route. The tool analyzes the topological properties of transport networks represented in the form of graphs. Large and complex graphs are broken down into hierarchies (subgraphs) to simplify the problem. First, we search for the shortest path between hierarchies, and only then, based on the selected subgraph, an accurate solution is found. By limiting the number of route options to go through, we greatly speed up the calculation process. This is especially important for logistics tasks that require operational calculations, such as, for example, organizing courier delivery of products, "said Georgy Kontsevik, a graduate student at ITMO and an employee of the Strong AI in Industry Research Center.

RIDE has been tested on graphs of street road networks in more than 500 cities. According to the test results, the hierarchical approach accelerates routing up to 30 times. At the same time, the error of calculations without using significant amounts of memory will be up to 10%, and if you use the system memory as much as possible, the error will be zero. Also, tests have shown that RIDE allows you to repeatedly form routes in a short time and is suitable for complex types of deliveries with many restrictions, for example, for transporting petroleum products by road. transport

Links

Library Reference