Physics-Informed Neural Network-Based Distributed Model Predictive Control for Multi-Vehicle Coordinated Pollination in Hybrid Rice Seed Production

Abstract

Abstract. Pollination is a critical process in hybrid rice seed production. However, traditional pollination methods suffer from inefficiency and the absence of a precise operational coordination control system. To address these limitations, this study proposes a framework integrating Physics-Informed Neural Networks (PINNs) with Distributed Model Predictive Control (DMPC) to rapidly compute optimization solutions, enabling coordinated pollination with multiple vehicles. Compared to existing model-based algorithms, the proposed method demonstrates superior performance in path tracking accuracy and formation coordination consistency during pollination. DMPC assigns an independent controller to each vehicle in the agricultural fleet, effectively handling complex system constraints and nonlinearities. Meanwhile, PINNs enhance prediction accuracy without requiring extensive training data by incorporating control information into the network through partial differential equations (PDEs). Experimental results show that the proposed method outperforms traditional Adaptive-PID and Pure-pursuit algorithms in tracking error reduction and asymptotic convergence, while maintaining formation consistency under dynamic speed adjustments. This research advances the precision of collaborative control in rice pollination, providing a foundation for modernizing hybrid rice seed production.

Abstract from the paper’s OpenAlex record, shown for indexing and research. Copyright remains with the authors or the publisher. Read at the source ↗

The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.