Global feasible path planning for pest monitoring robots in unstructured agricultural environments

In the context of sustainable agriculture and precision agriculture, autonomous mobile robots play a pivotal role in intelligent plant pest and disease detection. However, navigation within complex agricultural environments-characterized by narrow crop rows and irregular obstacles-remains a persistent challenge. To address these limitations, this study presents a Goal-Oriented Adaptive Bidirectional RRT* (GABi-RRT*) algorithm designed to overcome the issues of low sampling efficiency and unstable path quality that are inherent in the conventional RRT* algorithm in narrow spaces. The proposed GABi-RRT* algorithm introduces several key innovations to improve its performance in complex environments. In the sampling phase, an innovative dynamic goal-oriented probabilistic sampling strategy is introduced. This strategy adaptively adjusts the sampling probability throughout the planning process, taking into account the distribution of obstacles between the current node and the target. This approach significantly enhances the efficiency of the sampling process compared to traditional random sampling methods. In the RRT* expansion phase, the algorithm integrates the RRT* expansion mechanism with an enhanced Artificial Potential Field (APF) and introduces an adaptive step-size strategy. This combination improves the algorithm's search efficiency and enhances its exploratory capabilities, allowing for better navigation in diverse and challenging environments. For path optimization, the algorithm employs a hybrid approach that combines pruning optimization with cubic B-spline curve fitting. This method eliminates redundant nodes and smooths the generated path, thereby improving path quality and reducing the turning frequency of the monitoring robots. Finally, comparative experiments in a continuous narrow environment simulating crop rows reveal that the proposed GABi-RRT* algorithm outperforms several existing algorithms, including Bias-RRT*, Informed-RRT*, and P-RRT*. Specifically, the GABi-RRT* algorithm reduces the average running time by 45.39%, 49.71%, and 71.52%, respectively, and shortens the path length by 5.70%, 3.97%, and 1.60%. These results demonstrate the superior capabilities of the GABi-RRT* algorithm in terms of path quality, stability, and search efficiency, making it a promising solution for autonomous navigation in agricultural environments.