At present, there is extensive ongoing research and development on autonomous road vehicles, but much less effort is put on autonomous driving in off-road environments, especially with regard to autonomous machines driving in forest terrain. At the same time, forest machine work is harsh for the operator, and the operator’s maximum performance is a major bottleneck for machine operations. One way to enable a safe work environment is to remove the driver from the machine. Thus, forestry machines must become autonomous in order to improve personnel safety and work efficiency. The basis of enabling automation of forest machinery is highly precise and accurate acquisition of working environment information, especially reliable virtual representation of forest terrain and obstacles within. In this paper, the objective is to improve the accuracy of object detection in forest terrain based on an AI algorithm by adding the depth variation texture to corresponding area of RGB images. The results show that the algorithm significantly improves the accuracy of object detection while the probability of false detection is reduced. Meanwhile, computational efficiency compared to previous image processing is not reduced to such extent that it prevents real-time computation.
Forest Terrain Object Detection Based on RGB and Depth Information
2021-01-01
Conference paper
Electronic Resource
English
DDC: | 629 |
A Technique of Object Detection Using Information about Changes in the Terrain Image
Springer Verlag | 2017
|A Technique of Object Detection Using Information about Changes in the Terrain Image
Online Contents | 2017
|Combining depth and thermal information for object detection and avoidance
European Patent Office | 2018
|