JPDS-NN: Reinforcement Learning-Based Dynamic Task Allocation for Agricultural Vehicle Routing Optimization

2025-03-04

AI Summary

This paper addresses the Entry-Dependent Vehicle Routing Problem (EDVRP) in irregular agricultural fields, where route planning must jointly optimize multiple parameters while respecting field geometry and entry-point constraints. We propose an end-to-end learning framework that integrates geometric awareness with dynamic scheduling. Unlike conventional heuristic methods that ignore field shape and entry restrictions, our approach introduces a novel graph neural network architecture based on joint probability distribution sampling, explicitly encoding field geometry and entry constraints as a structured graph. The framework incorporates a Graph Transformer encoder, cross-attention mechanisms, and Proximal Policy Optimization (PPO) for reinforcement learning. It supports both pre-training and online fine-tuning, enabling dynamic task allocation and real-time re-planning. Experiments demonstrate significant improvements over baseline methods: travel distance reduced by 48.4–65.4%, fuel consumption decreased by 14.0–17.6%, computational efficiency improved by two orders of magnitude, and dynamic scheduling performance enhanced by 15–25%.

Technology Category

• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

The Entrance Dependent Vehicle Routing Problem (EDVRP) is a variant of the Vehicle Routing Problem (VRP) where the scale of cities influences routing outcomes, necessitating consideration of their entrances. This paper addresses EDVRP in agriculture, focusing on multi-parameter vehicle planning for irregularly shaped fields. To address the limitations of traditional methods, such as heuristic approaches, which often overlook field geometry and entrance constraints, we propose a Joint Probability Distribution Sampling Neural Network (JPDS-NN) to effectively solve the EDVRP. The network uses an encoder-decoder architecture with graph transformers and attention mechanisms to model routing as a Markov Decision Process, and is trained via reinforcement learning for efficient and rapid end-to-end planning. Experimental results indicate that JPDS-NN reduces travel distances by 48.4-65.4%, lowers fuel consumption by 14.0-17.6%, and computes two orders of magnitude faster than baseline methods, while demonstrating 15-25% superior performance in dynamic arrangement scenarios. Ablation studies validate the necessity of cross-attention and pre-training. The framework enables scalable, intelligent routing for large-scale farming under dynamic constraints.

Problem

• Optimizes agricultural vehicle routing for irregular fields
• Addresses limitations of traditional heuristic methods
• Enhances efficiency with reinforcement learning-based JPDS-NN

Innovation

• JPDS-NN uses reinforcement learning for dynamic task allocation.
• Encoder-decoder architecture with graph transformers enhances routing.
• Attention mechanisms optimize agricultural vehicle routing efficiency.