The world's first Autonomous drone navigation system for technical inspections inside structures

The technology is based on computer vision and the method of simultaneous localization and map construction (SLAM).
Localization, mapping, detection and identification of objects using proprietary computer vision algorithms, deep convolutional neural networks and a combination of a set of sensors installed on the UAV.
Our product is a platform for developing applications for industrial drone applications.

Our technology does not require pre-preparation of the map. The drone is equipped with a variety of sensors that help it determine the location, tilt, altitude and other important parameters for the flight. At the moment, our navigation system consists of the following sensors: scanning circular lidar, laser rangefinders, stereo camera, Jetson TX2, IMU, flashlight and software. Special algorithms allow the drone to identify the position in space based on the collected information and control the movement independently.

On our platform, we can develop a navigation system for similar types of robots, such as crawlers.

The difference between our technology and that of our competitors. First, our rather complex system on SLAM starts with a single button. Ideally, you can put the drone, press the button, move away, the drone will take off and complete the mission and land in the same place. After that, the person can turn off the drone, move it to another location, and conduct the next inspection. The technology is really Autonomous, it does not require human piloting skills.

Second, our navigation system consists of several components. - this is visual odometry, this is navigation based on laser scanners-lidar. General information is processed and synthesized on the basis of all these systems. This means that if the drone falls into conditions where one of the systems does not work, the drone continues to fly based on the other. For example, if a drone enters a complex environment where there are many wires and pipes, it continues to fly based on computer vision. We have three levels of navigation that are present there. Knocks one off, continues to fly to the other as conditions change. ‍

Using a drone to survey oil storage facilities, you can perform a full survey, build a three-dimensional model and assess the condition of the tank, almost without entering it.

The drone on such a platform will be able to serve any tasks involving internal navigation and indoor flights in the absence of lighting, GPS signals, radio signals, or incorrect operation of the magnetometer.

Limiting the risks of drift collapse is a core concern for mining companies. Landslides are especially frequent after technical explosions, which are an essential part of field development. Relying just on its sensors our drone can fly autonomously through the drift and build a 3D map of the area. It does this by using a 3D laser scanner to perform a photo shoot mapping all the inner spaces.
Based on this map, the mining crew’s decisions are far more accurate, which means a much safer situation for workers, since they would know the location of potentially dangerous places before moving into the area.