Evolving Edge Intelligence for Adaptive and Autonomous Smart Agriculture

Special Session at the 2026 IEEE International Conference on Evolving and Adaptive Intelligent Systems (IEEE EAIS 2026)

Location : Pisa, Italy

Date : September 21-23, 2026

AIM and Scope

Modern agriculture is a paradigmatic non-stationary data-stream environment. Seasonality, weather, crop variability, soil heterogeneity, and machinery degradation over time affect both actuation and sensing quality. These factors continuously reshape data distributions and operating conditions, demanding evolving and adaptive intelligent systems that can remain reliable after deployment. In this context, the next step beyond large-scale sensing is closed-loop intelligence, in which multimodal perception is directly coupled with reasoning and autonomous actuation, and models are continuously monitored and improved throughout the season.

This special session focuses on closed-loop, field-deployable Edge AI for precision and smart agriculture. It integrates heterogeneous sensors, ranging from IoT to computer vision and LiDAR, for SLAM, and includes actuators such as autonomous tractors, implementing agricultural robotics for scouting, weeding, spraying, and harvesting. We emphasize EAIS challenges central to real deployments, including concept drift in streaming data, intermittent connectivity with offline-first operation, strict compute and power budgets, and harsh environmental conditions. Contributions are encouraged that address adaptive control and decision systems, robust and interpretable online inference, and self-monitoring capabilities to detect degradation, uncertainty, and out-of-distribution situations.

A key focus is the operationalization of evolving models through MLOps and TinyMLOps. This includes monitoring device performance, data, and model versioning, safe and controllable updates, including reprogrammability while systems are deployed, and resilient edge-cloud cooperation. We welcome works that combine methodological advances with measurable agronomic and operational impact, validated on realistic data streams and in laboratory or field settings, including autonomous UGV and UAV platforms and agricultural machinery.

Session topics

Main topics related to this special session include, but are not limited to:

Sensing-to-action closed-loop architectures for Smart/Precision Agriculture
Agricultural MLOps: deployment, monitoring, drift, retraining, safe OTA updates
Edge AI: in-field/on-device inference and edge-cloud collaboration
Continual/online learning under seasonality and concept drift
Robust perception in harsh/unstructured farm conditions (vision/LiDAR, SLAM)
Detection/segmentation/tracking, sensor fusion, and fault-tolerant perception
Autonomous agricultural robotics (UGV/UAV): scouting, mapping, precision intervention; GNSS-denied navigation (orchards/vineyards)
Adaptive control and decision systems for tractor/UGV/UAV platforms
Decision support & optimization: irrigation, soil moisture, stress/disease, yield/risk forecasting
Distributed/federated and privacy-preserving learning across farms/devices
Explainable and uncertainty-aware AI for farmers and agronomists
Simulation and synthetic data for training and validation
Validation & benchmarking with measurable agronomic/operational outcomes in realistic field trials