Low-Cost Plant-Protection Unmanned Ground Vehicle System for Variable Weeding Using Machine Vision

Huangtao Dong, Jianxun Shen, Zhe Yu, Xiangyu Lǚ, Fei Liu, Wenwen KongSensors 2024 · DOI

TL;DR

This paper introduces a low-cost, machine vision-guided Unmanned Ground Vehicle (UGV) system for precise, variable-rate weeding. By combining advanced image processing with an adaptive fuzzy PID control system, it significantly reduces pesticide waste and environmental impact compared to traditional methods.

What Problem Does This Paper Address?

Traditional agricultural spraying methods often lead to excessive pesticide use and environmental pollution due to their uniform application across fields. This indiscriminate spraying results in wasted chemicals, potential health risks, and soil contamination. The paper addresses the need for a more sustainable and efficient approach by developing a system that can precisely target weeds, applying pesticides only where and when needed. This "on-demand" spraying aims to minimize chemical usage while maintaining crop yield and protecting the environment.

Methods

The core of the system is a machine vision pipeline integrated with a variable spraying control subsystem on an UGV.

  • Machine Vision: The system employs a visible light camera to capture images of vegetation. These images are first converted to grayscale and then processed using a combination of the normalized super green algorithm (2G-R-B) for enhanced vegetation contrast and the Fast Iterative Threshold Segmentation (FIT) method for accurate segmentation of vegetation from the background. This allows for precise identification and mapping of weed distribution.
  • Adaptive Fuzzy PID Control: To ensure stable and accurate spraying, the system utilizes an adaptive fuzzy PID controller. This controller dynamically adjusts the PID parameters (Kp, Ki, Kd) based on fuzzy logic rules to manage water pressure fluctuations in the spray system. This is crucial for maintaining consistent spray output, especially under varying conditions.
  • Staggered-Period PWM Control: To further mitigate pressure fluctuations and improve spraying accuracy, a staggered-period Pulse Width Modulation (PWM) control strategy is implemented. This method sequences the activation of individual spray nozzles, preventing simultaneous operation and thus reducing pressure spikes.
  • UGV Integration: The system is integrated with an UGV that also features autonomous path planning capabilities, enabling fully automated variable weeding operations.

Datasets

The paper does not explicitly mention training datasets. However, it utilizes a collection of 40 representative vegetation images captured from field experiments for testing and evaluating the image segmentation algorithms.

Key Insights and Ideas

  • The Power of Combined Algorithms: The success of the system hinges on the synergistic combination of algorithms. The 2G-R-B normalization and FIT segmentation proved to be a robust pairing for distinguishing vegetation, while the adaptive fuzzy PID controller effectively tackled the inherent pressure instability in variable spraying.
  • Fuzzy Logic for Real-World Chaos: Applying fuzzy logic to tune PID parameters is a smart way to handle the non-linear and unpredictable nature of agricultural environments, where factors like terrain and plant density constantly change.
  • Staggering for Stability: The staggered PWM approach is a clever, low-cost solution to a significant engineering challenge in variable spraying – maintaining consistent pressure. It highlights how simple timing adjustments can yield substantial improvements in system performance.
  • Low-Cost, High-Impact: The emphasis on "low-cost" components and algorithms (visible light camera, standard microcontrollers) makes this technology more accessible for widespread adoption, addressing a key barrier to precision agriculture.

Key Results

  • Pressure Stability: The adaptive fuzzy PID control system achieved a pressure variation of less than ±0.05 MPa.
  • Spraying Accuracy: The average spraying error was consistently below 2%, with the highest error not exceeding 5% across various duty cycles.
  • Image Segmentation: The combination of 2G-R-B and FIT algorithms demonstrated efficient and accurate differentiation between vegetation and background.
  • Response Time: The overall system response time, from image acquisition to spraying, was approximately 0.0482 seconds, which is sufficient for real-time operation.
  • Field Trial Success: Field trials confirmed the system's ability to adjust spraying volume in real-time based on vegetation distribution, proving its effectiveness and low cost.

Limitations and Future Work

  • Image Acquisition Interval: The current image acquisition interval is 2 seconds, which might still be a limitation for very fast-moving UGVs or rapidly changing weed growth.
  • UGV Speed Variation: While the system now accounts for UGV speed variations using a Hall sensor, the paper notes that unpredictable factors like slope and humidity can still interfere with travel speed, potentially causing discontinuities in image capture.
  • Future Work: The authors suggest that the system's adaptability to different carriers and its potential for industrialization are promising avenues for future development. Further research could focus on optimizing image acquisition speed and robustness against more complex environmental variations.

Summary updated

Abstract

This study presents a machine vision-based variable weeding system for plant- protection unmanned ground vehicles (UGVs) to address the issues of pesticide waste and environmental pollution that are readily caused by traditional spraying agricultural machinery. The system utilizes fuzzy rules to achieve adaptive modification of the Kp, Ki, and Kd adjustment parameters of the PID control algorithm and combines them with an interleaved period PWM controller to reduce the impact of nonlinear variations in water pressure on the performance of the system, and to improve the stability and control accuracy of the system. After testing various image threshold segmentation and image graying algorithms, the normalized super green algorithm (2G-R-B) and the fast iterative threshold segmentation method were adopted as the best combination. This combination effectively distinguished between the vegetation and the background, and thus improved the accuracy of the pixel extraction algorithm for vegetation distribution. The results of orthogonal testing by selected four representative spraying duty cycles-25%, 50%, 75%, and 100%-showed that the pressure variation was less than 0.05 MPa, the average spraying error was less than 2%, and the highest error was less than 5% throughout the test. Finally, the performance of the system was comprehensively evaluated through field trials. The evaluation showed that the system was able to adjust the corresponding spraying volume in real time according to the vegetation distribution under the decision-making based on machine vision algorithms, which proved the low cost and effectiveness of the designed variable weed control system.
BibTeX
@inproceedings{dong2024low,
  author = {Huangtao Dong and Jianxun Shen and Zhe Yu and Xiangyu Lǚ and Fei Liu and Wenwen Kong},
  title = {{Low-Cost Plant-Protection Unmanned Ground Vehicle System for Variable Weeding Using Machine Vision}},
  booktitle = {Sensors},
  year = {2024},
  doi = {10.3390/s24041287},
}
Figure