Developers introduced the jev-drone drone control system

The development team has released jev-drone — a system for autonomous control of drones using a specialized decision-making model. The project is focused on performing complex tactical maneuvers in a dynamic environment.

The system is based on the TypeSafe judgment model, which operates at a frequency of 2.5 Hz. The model is responsible for high-level tactical decisions, such as gaining altitude or flying through narrow obstacles. A separate geometric controller working in conjunction with the logical core is used to ensure flight safety and stabilization.

The effectiveness of the architecture was confirmed during testing on a 5-station obstacle course. The drone successfully completed the route, demonstrating its ability to autonomously navigate in a space with a high density of objects. The project code is available in an open repository under a license that allows integration into third-party control systems.

The use of the TypeSafe approach allows minimizing errors when interpreting sensor data, which is critical for autonomous flights. Unlike standard neural network solutions, this model focuses on strict adherence to given safety parameters when performing maneuvers.

The development represents an example of optimizing computational resources for robotics tasks. The update frequency of 2.5 Hz was chosen as a balance between the speed of response to obstacles and the computational load on the onboard controller, allowing the system to be used on compact devices with limited power consumption.