Obstacle identification and classification method utilizing laser displacement sensors

This study introduces an advanced obstacle detection and recognition approach for plant protection unmanned aerial vehicles (UAVs), leveraging laser displacement sensor technology. The proposed methodology encompasses three core processes: extraction of data segments, computation of fundamental obstacle parameters, and classification through pattern recognition. By evaluating the reliability of distance measurements, relevant data blocks corresponding to potential obstacles are isolated from the sequential sensor outputs. These obstacles are then characterized using their average angular position, mean distance, and spatial width. A pattern recognition algorithm is developed based on four distinctive features: obstacle width, maximum interval within the data block, the frequency of abrupt data changes, and variance. Experimental validation demonstrates that the method reliably identifies both the orientation and distance of obstacles in complex and unfamiliar environments, while effectively distinguishing between different obstacle types.