Source-linked AI summary
Utility-Driven Spatial Data Sampling for UAV-Assisted Scientific Smart Farming
Keiwan Soltani, Sajal K. Das
TL;DR
Energy-constrained UAVs cannot collect every sensor reading on each mission, and common collection objectives do not distinguish scientific value from data availability. The framework uses grid-cell utility to select a subset of sites under battery and return constraints. In a preliminary semi-synthetic deployment, it selected eight cells and returned safely within the usable battery budget.
Problem
Energy-constrained UAVs cannot collect all sensor data during every mission, and existing collection objectives do not always distinguish merely available data from scientifically informative data.
Source: §1 · p. 1Method
Results
55,309.8 J total mission energy was below the 66,500 J usable battery budget (83.2% utilization); the UAV selected eight cells and returned safely.
Source: §5 · p. 2Takeaways & Limitations
Preliminary results demonstrate that the framework can prioritize high-value cells while respecting the UAV energy constraint.
Source: §5 · p. 2
Abstract
from arXiv · showhide
Large smart-farming deployments generate continuous scientific data from spatially distributed sensors, including soil, humidity, temperature, crop-health, and pest-related measurements. In vast agricultural fields, however, an energy-constrained unmanned aerial vehicle (UAV) often cannot collect data from every sensor during each mission. Existing UAV-assisted collection methods typically optimize coverage, route length, data volume, or freshness, but they do not always distinguish between data that is merely available and data that is scientifically valuable. This poster introduces a utility-driven spatial sampling framework for UAV-assisted smart farming. The field is partitioned into grid cells, each sized according to the UAV ground coverage range. After an initial exploration phase, each cell receives a scientific utility score based on freshness, redundancy, anomaly likelihood, and model uncertainty. The UAV then selects and visits a subset of high-utility cells under battery and return-to-base constraints. The proposed framework reframes UAV-based collection as adaptive scientific data management rather than exhaustive sensing.
1 Introduction
Energy-constrained UAVs cannot collect all sensor data on every mission, while common collection objectives do not distinguish merely available data from scientifically informative data.
1 Introduction
- Large, connectivity-limited farms can exceed an energy-constrained UAV’s capacity to collect all sensor data in one mission.Smart farms generate continuous soil, environmental, crop-health, and pest-related measurements.Source: §1 · p. 1
- Existing UAV collection methods commonly optimize trajectory length, coverage, or freshness without distinguishing scientifically informative data from merely available data.Source: §1 · p. 1
- The proposed framework scores each UAV-coverage cell by freshness, anomaly likelihood, model uncertainty, and redundancy, then selects an energy-feasible subset of high-value cells.Source: §1 · p. 1
2 System Overview
The system groups sensors into UAV-coverage grid cells and proceeds from broad initial exploration to selective, utility-guided collection.
2 System Overview
- The field is divided into grid cells sized to the UAV’s ground coverage, allowing the UAV to collect from sensors within a cell when it reaches its center.Collection within the cell uses Orthogonal Frequency-Division Multiple Access (OFDMA).Source: §2 · p. 1
- During one or more early flights, the UAV explores all cells to build a baseline spatial profile of sensor behavior and field patterns.The profile includes normal behavior, temporal patterns, spatial correlations, historical anomaly levels, and model performance across regions.Source: §2 · p. 1
- In adaptive collection, the UAV assigns each cell a scientific utility score and plans an energy-feasible route through a selected subset rather than visiting every cell.Source: §2 · p. 1
- A cell-level metadata table stores collection times, summaries, nearby-cell correlations, anomaly history, model uncertainty, estimated data volume, and expected collection energy.Mission planning uses this metadata before departure and can be updated after each mission.Source: §2 · p. 1
3 Cell-Level Scientific Utility
The framework assigns each grid cell a scientific utility score from freshness, redundancy, anomaly likelihood, and model uncertainty, reflecting how valuable collection is for the current mission.
3 Cell-Level Scientific Utility
- Each cell receives a utility score estimating its collection value for the current mission, based on freshness, redundancy, anomaly, and model uncertainty.Source: §3 · p. 1
- Cell-level time series are formed by aggregating sensor measurements within each cell for each sensing modality.Examples of modalities include soil moisture, temperature, humidity, and water level.Source: §3 · p. 1
- Redundancy estimates average positive historical correlation between a cell’s aggregated series and those of neighboring cells; highly similar cells may be less urgent to collect.Source: §3 · p. 1
- Anomaly likelihood combines modality-specific anomaly scores to identify cells whose current aggregated measurements are unusual relative to their historical patterns.Such patterns may reflect localized water stress, irrigation irregularity, environmental change, or another agricultural event.Source: §3 · p. 2
- Model uncertainty is measured with normalized Shannon entropy over predicted agricultural-condition classes; high uncertainty indicates the model cannot confidently determine a cell’s condition.Collecting data from such a cell may be useful for updating the downstream monitoring model.Source: §3 · p. 2
- Application-specific weights combine the dimensions so the utility formulation rewards stale, anomalous, or uncertain cells and penalizes cells whose data are likely redundant.Source: §3 · p. 2
4 Energy-Aware Route Selection
Route selection maximizes collected cell utility under an energy budget, with travel, hovering, communication, and safe return included in the feasibility constraint.
4 Energy-Aware Route Selection
- The UAV selects and orders a subset of cells to maximize total collected utility under its battery limit, making the problem closer to orienteering than a Traveling Salesman Problem.Each candidate location has a reward and the route is constrained by a travel budget.Source: §4 · p. 2
- Route feasibility accounts for travel, hovering, communication and data reception, and the energy needed to return safely to base within the battery budget.Source: §4 · p. 2
- A practical heuristic selects the next cell using marginal utility per additional route energy, admitting it only when collection and return remain feasible.Source: §4 · p. 2
5 Preliminary Evaluation
A preliminary evaluation on a field-scale soil-moisture dataset used a semi-synthetic spatial deployment. The utility-aware route selected eight cells and returned safely within the usable battery budget.
5 Preliminary Evaluation
- 55,309.8 J of mission energy was 83.2% of the 66,500 J usable battery budget for the eight-cell route, which returned safely to base.Route planning accounted for travel, hovering, communication, and return-to-base energy.Source: §5 · p. 2
- The utility-aware route prioritized high-utility cells while accounting for travel, hovering, communication, and return-to-base energy.The field was divided into UAV-coverage cells scored for freshness, anomaly likelihood, uncertainty, and redundancy.Source: §5 · p. 2
- The framework was evaluated on a field-scale soil-moisture sensor-network dataset under a semi-synthetic spatial deployment.Source: §5 · p. 2
- The preliminary results show that the framework can prioritize high-value cells while respecting the UAV energy constraint.Source: §5 · p. 2