Obstacle avoidance algorithms, SLAM, stereovision
Drone machine vision: simultaneous localization and mapping (SLAM)
We developed an optical navigation system based on drone machine vision SLAM algorithms (relative optical navigation), Algorithm of Morphological Correlation-Extreme Navigation System. It is an absolute navigation system for a preloaded map and IMU. All these three information layers were combined by a single filter, taking into account the budget of errors and the time of reliability of the data of each information layer. This filter actually made a decision about the actual position of the object.
We also improved the stability of MEMS IMU readings by using redundant navigation systems in non-orthogonal positions. This allowed obtaining high stability indicators for household IMU.
Drone machine vision, according to the conditions of the task, had to determine its location no worse than 2 meters after it had flown 70 kilometers without the use of GPS.
Interested in how new technologies could help your business? Check how it works for agriculture: Insects sensor array for precision agriculture