Orchard local sparse mapping method and system based on binocular vision and RTK

Abstract

The invention discloses an orchard local sparse mapping method and system based on binocular vision and RTK, and the method comprises the steps: recognizing a fruit tree based on a deep learning method through employing a high-precision positioning RTK and a binocular vision distance measurement module installed by an agricultural operation vehicle, obtaining the longitude and latitude coordinatesof each tree through binocular distance measurement in combination with the RTK, and obtaining local sparse mapping. The system disclosed by the invention comprises a binocular vision module and an RTK positioning module, the binocular vision module comprises a left camera and a right camera which are used for collecting orchard fruit tree vision images, and the left and right vision images are calculated to obtain fruit tree coordinates; the RTK positioning module acquires longitude and latitude information of the position where the RTK module is located; the binocular vision module and theRTK positioning module are installed on an operating vehicle, and a left camera and a right camera of the binocular vision module are symmetrically installed on the left side and the right side of anRTK antenna of the RTK positioning module.

Description

technical field [0001] The invention relates to an orchard mapping method and system, which finally realizes local sparse mapping of the orchard by acquiring the latitude and longitude coordinates of each fruit tree. Background technique [0002] Maps are the premise of path planning for unmanned agricultural machinery operations in the agricultural field. Especially, automatic navigation and driving in mountainous and hilly areas in my country are inseparable from maps of the operating environment. Due to the lack of operating environment maps, the current situation has been limited to a certain extent. The practical application and promotion of automatic driving operations of unmanned agricultural machinery in my country. At present, the method based on path planning, whether it is the traditional artificial potential field method, ant colony algorithm or A*, D* algorithm, or the popular SLAM (real-time positioning and mapping) or VSLAM (vision-based real-time positioning a...

AI Technical Summary

Problems solved by technology

Due to the particularity of the orchard environment, currently commonly used mapping methods such as SLAM or UAV aerial photography cannot effectively solve the path planning problem of unmanned walking machines

For example, SLAM is mostly used indoors. In outdoor environments, visual SLAM is easily affected by light conditions. Laser SLAM is expensive, and single-line lidar cannot meet the needs at all.

Although UAV aerial photography can build a relatively high-precision top-view map, the labor cost is high, and professional personnel are required to take aerial photography and construct the map. If the environment changes, the map must be rebuilt. In addition, the coordinates of the map and the actual positioning are required. Matching, it is difficult for agricultural machinery operators or managers to operate technically

Images