Russian Researchers Find Way to Speed Up AI Navigation 100-Fold
A new system developed by Russian scientists can accelerate AI-based navigation by up to 100 times and is already ready for real-world deployment.
Russian researchers have developed a new approach that could significantly speed up AI navigation, potentially boosting the development of autonomous transport systems.
The project involved multiple research teams, including the Moscow Institute of Physics and Technology, the Ufa University of Science and Technology, and a cognitive AI systems laboratory. The method is based on visibility graphs, which represent space through angles and the connections between them. A key improvement is the vectorization of computations, allowing the system to replace slower pairwise checks with batch data processing. Another innovation is a geometry simplification system for obstacles, which reduces computational load.
According to the developers, the system has already been tested on map models and is now ready for full-scale deployment in robotic warehouses, search-and-rescue systems, and autonomous drone navigation.
