Srivani et al. (2026) Revolutionizing Agriculture: Smart Irrigation 4.0 with Reinforcement Learning Using Deep Q-Network

Identification

• Journal: Communications in computer and information science
• Year: 2026
• Date: 2026-01-01
• Authors: M. Srivani, G. Dinesh, S. A. Athi Lakshmi
• DOI: 10.1007/978-3-032-14531-4_19

Research Groups

Department of AI and ML, Sri Ramachandra Faculty of Engineering and Technology, SRIHER, Porur, Chennai, Tamil Nadu, India

Short Summary

This study proposes a smart irrigation system utilizing a Deep Q-Network (DQN) to optimize water efficiency by maintaining soil moisture between 30% and 50%. The system demonstrates a 30% reduction in water wastage and achieves optimal moisture levels in 85% of cases compared to conventional rule-based methods.

Objective

• To develop and implement a smart irrigation system using reinforcement learning (Deep Q-Network) to optimize water efficiency by maintaining soil moisture levels between 30% and 50%.

Study Configuration

• Spatial Scale: Simulated agricultural environment
• Temporal Scale: Simulated over numerous episodes

Methodology and Data

• Models used: Reinforcement Learning (RL), Deep Q-Network (DQN) algorithm
• Data sources: Real-time weather data from OpenWeatherMap API (temperature, humidity, precipitation), simulated soil moisture dynamics

Main Results

• The proposed RL-based method reduces water wastage by 30% compared to conventional rule-based systems.
• Optimal soil moisture levels are maintained in 85% of cases.
• The system achieved an average reward of 4996.8 with a standard deviation of 6.4, indicating consistent performance.
• It effectively reduces both over and under-irrigation, ensuring precise water management.

Contributions

• Introduction of an adaptive, reinforcement learning-based (DQN) smart irrigation system that significantly outperforms conventional rule-based systems in water efficiency and soil moisture regulation.
• The system demonstrates the ability to adapt to changing environmental conditions, a limitation of traditional rule-based approaches.
• Provides a scalable and dependable solution for sustainable agricultural practices.

Funding

Not specified in the provided text.

Citation

@article{Srivani2026Revolutionizing,
  author = {Srivani, M. and Dinesh, G. and Lakshmi, S. A. Athi},
  title = {Revolutionizing Agriculture: Smart Irrigation 4.0 with Reinforcement Learning Using Deep Q-Network},
  journal = {Communications in computer and information science},
  year = {2026},
  doi = {10.1007/978-3-032-14531-4_19},
  url = {https://doi.org/10.1007/978-3-032-14531-4_19}
}