Design and Testing of Bionic-Feature-Based 3D-Printed Flexible End-Effectors for Picking Horn Peppers

Deng, Lexing; Liu, Tianyu; Jiang, Ping; Qi, Aolin; He, Yuchen; Li, Yujie; Yang, Mingqin; Deng, Xin

doi:10.3390/agronomy13092231

Cited by 5 publications

(2 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Picking methods based on intelligent algorithms have high picking accuracies. Deng et al designed a flexible and effective method for chili picking [5]. The drum snap finger picking device has the advantages of picking flexibility and high variety adaptability.…”

Section: Introductionmentioning

confidence: 99%

Design and Experimental Study of a Biomimetic Pod-Pepper-Picking Drum Based on Multi-Finger Collaboration

Du,

Fang,

Han

et al. 2024

Agriculture

View full text Add to dashboard Cite

In order to reduce ground drop loss during mechanical pepper picking and improve the net recovery rate, a drum snap finger picking device was designed. The picking device is mainly composed of a picking drum and auxiliary picking components; the picking finger arrangement was designed biomimetically and its structure and operating parameters were optimized by the DEM (discrete element method). According to the physical and mechanical characteristics of the pepper and the simplified three-dimensional model of the picking device, a virtual simulation model of the pepper-picking device was established using the EDEM software. Through simulation analysis and using the orthogonal test method, the main factors which affect the ground drop loss rate of pepper and their optimal parameter combination values were determined. The simulation results were verified by a pepper-picking field experiment. Orthogonal tests show that, when the picking drum speed (V′) is 210 rpm, the pepper-feeding speed (V″) is 1100 mm·s−1, the bending angle of each picking spring tooth (C) is 162°, and each group of circumferential fingers has rows, the picking device has a good picking effect. At this time, the ground drop loss rates in both the simulation and field test were 7.50% and 7.85%, respectively, and the drop error was only 4.46%, which was within the allowable range. The design form and parameter optimization simulation method in this paper provide an important reference for the design and optimization of pepper-harvesting machinery.

show abstract

Section: Introductionmentioning

confidence: 99%

Design and Experimental Study of a Biomimetic Pod-Pepper-Picking Drum Based on Multi-Finger Collaboration

Du,

Fang,

Han

et al. 2024

Agriculture

View full text Add to dashboard Cite

show abstract

“…These results indicated that our improvements were very effective in enhancing the detection performance of the proposed model. In addition, our improved algorithm is well suited for edge deployment and embedded development with the help of some control algorithms [43] and hardware device [44], as the inference speed of the model ensures the feasibility of real-time detection. 3.…”

mentioning

confidence: 99%

A Forest Wildlife Detection Algorithm Based on Improved YOLOv5s

Yang,

Liu,

Jiang

et al. 2023

Animals

Self Cite

View full text Add to dashboard Cite

A forest wildlife detection algorithm based on an improved YOLOv5s network model is proposed to advance forest wildlife monitoring and improve detection accuracy in complex forest environments. This research utilizes a data set from the Hunan Hupingshan National Nature Reserve in China, to which data augmentation and expansion methods are applied to extensively train the proposed model. To enhance the feature extraction ability of the proposed model, a weighted channel stitching method based on channel attention is introduced. The Swin Transformer module is combined with a CNN network to add a Self-Attention mechanism, thus improving the perceptual field for feature extraction. Furthermore, a new loss function (DIOU_Loss) and an adaptive class suppression loss (L_BCE) are adopted to accelerate the model’s convergence speed, reduce false detections in confusing categories, and increase its accuracy. When comparing our improved algorithm with the original YOLOv5s network model under the same experimental conditions and data set, significant improvements are observed, in particular, the mean average precision (mAP) is increased from 72.6% to 89.4%, comprising an accuracy improvement of 16.8%. Our improved algorithm also outperforms popular target detection algorithms, including YOLOv5s, YOLOv3, RetinaNet, and Faster-RCNN. Our proposed improvement measures can well address the challenges posed by the low contrast between background and targets, as well as occlusion and overlap, in forest wildlife images captured by trap cameras. These measures provide practical solutions for enhanced forest wildlife protection and facilitate efficient data acquisition.

show abstract