The orchards usually have rough terrain，dense tree canopy and weeds. It is hard to use GNSS for autonomous navigation in orchard due to signal occlusion，multipath effect，and radio frequency interference. To achieve autonomous navigation in orchard，a visual navigation method based on multiple images at different shooting angles is proposed in this paper. A dynamic image capturing device is designed for camera installation and multiple images can be shot at different angles. Firstly，the obtained orchard images are classified into sky and soil detection stage. Each image is transformed to HSV space and initially segmented into sky，canopy and soil regions by median filtering and morphological processing. Secondly，the sky and soil regions are extracted by the maximum connected region algorithm，and the region edges are detected and filtered by the Canny operator. Thirdly，the navigation line in the current frame is extracted by fitting the region coordinate points. Then the dynamic weighted filtering algorithm is used to extract the navigation line for the soil and sky detection stage，respectively，and the navigation line for the sky detection stage is mirrored to the soil region. Finally，the Kalman filter algorithm is used to fuse and extract the final navigation path. The test results on 200 images show that the accuracy of visual navigation path fitting is 95. 5%，and single frame image processing costs 60 ms，which meets the real-time and robustness requirements of navigation. The visual navigation experiments in Camellia oleifera orchard show that when the driving speed is 0. 6 m/s，the maximum tracking offset of visual navigation in weed-free and weedy environments is 0. 14 m and 0. 24 m，respectively，and the RMSE is 30 mm and 55 mm，respectively.

In China, tea products made from fresh leaves characterized by one leaf with one bud (1L1B) are classified as “Famous Tea”, which has better taste and higher economic value, but suffers from a labor shortage. Aiming at picking automation, existing studies focus on visual detection of 1L1B, but algorithm validation is limited to a specific variety of tea sprouting in a certain harvest season at a certain location, which limits the engineering application of developed tea picking robots working in various natural tea fields. To address this gap, a deep learning model DMT (detecting multispecies of tea) based on YOLOX-S was proposed in this paper. The DMT network takes YOLOX-S as a baseline and adds ECA-Net to the CSP Darknet and FPN of YOLOX-S. The average precision (AP), precision, and recall of DMT are 94.23%, 93.39%, and 88.02%, respectively, for detecting 1L1B sprouting in spring; 93.92%, 93.56%, and 87.88%, respectively, for detecting 1L1Bsprouting in autumn. These experimental results are better than those of the five current object detection models. After fine-tuning the DMT network with another dataset composed of multiple tea varieties, the DMT network can detect 1L1B for different varieties of tea in multiple picking seasons. The results can promote the engineering application of picking automation of fresh tea leaves.