IntelliFly: Multi-UAV Cooperative Path Optimization for IoT-enabled Energy-efficient Smart Agriculture

Engineering notes

The integration of unmanned aerial vehicles (UAVs) with Internet of Things (IoT) systems plays a critical role in enabling real-time monitoring, efficient resource utilization, and energy-aware operations in precision agriculture.However, existing multi-UAV frameworks often exhibit limitations in energy-efficient path planning and lack effective multimodal alignment between UAV imagery and IoT sensor data, leading to reduced decision accuracy and suboptimal field coverage.To address these challenges, this study proposes IntelliFly, a cooperative multi-UAV framework that integrates Vision Transformer (ViT)-based feature extraction with a Transformer Fusion Network (TFN) for semantically consistent multimodal data fusion.In addition, a Particle Swarm Optimization (PSO)-based strategy is employed to achieve adaptive and energy-aware UAV path planning.The framework is implemented using TensorFlow and PyTorch and evaluated on the Semantic-Aware IoT Smart Agriculture dataset from Kaggle.To ensure experimental rigor, key baseline models are re-implemented under identical dataset splits, preprocessing procedures, and evaluation protocols.The experimental results from five independent runs show that the proposed IntelliFly framework achieves 96.2% accuracy, improving performance by 1.9 to 9.8 percentage points over the baseline methods.It also improves energy efficiency by 3.8 to 7.4 percentage points and reduces computation time by 25.8%, from 12.8 seconds to 9.5 seconds.Overall, the findings indicate that IntelliFly effectively integrates semantic perception with cooperative UAV optimization, enabling enhanced coverage and energy-aware coordination.The proposed framework provides a reproducible and scalable solution for multimodal UAV-IoT systems and supports the development of intelligent, real-time agricultural monitoring applications.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

The integration of unmanned aerial vehicles (UAVs) with Internet of Things (IoT) systems plays a critical role in enabling real-time monitoring, efficient resource utilization, and energy-aware operations in precision agriculture.However, existing multi-UAV frameworks often exhibit limitations in energy-efficient path planning and lack effective multimodal alignment between UAV imagery and IoT sensor data, leading to reduced decision accuracy and suboptimal field coverage.To address these challenges, this study proposes IntelliFly, a cooperative multi-UAV framework that integrates Vision Transformer (ViT)-based feature extraction with a Transformer Fusion Network (TFN) for semantically consistent multimodal data fusion.In addition, a Particle Swarm Optimization (PSO)-based strategy is employed to achieve adaptive and energy-aware UAV path planning.The framework is implemented using TensorFlow and PyTorch and evaluated on the Semantic-Aware IoT Smart Agriculture dataset from Kaggle.To ensure experimental rigor, key baseline models are re-implemented under identical dataset splits, preprocessing procedures, and evaluation protocols.The experimental results from five independent runs show that the proposed IntelliFly framework achieves 96.2% accuracy, improving performance by 1.9 to 9.8 percentage points over the baseline methods.It also improves energy efficiency by 3.8 to 7.4 percentage points and reduces computation time by 25.8%, from 12.8 seconds to 9.5 seconds.Overall, the findings indicate that IntelliFly effectively integrates semantic perception with cooperative UAV optimization, enabling enhanced coverage and energy-aware coordination.The proposed framework provides a reproducible and scalable solution for multimodal UAV-IoT systems and supports the development of intelligent, real-time agricultural monitoring applications.

Links and sources

• PDF from original source

Comments