A drone that learns to efficiently find objects in agricultural fields: from simulation to the real world

2025-05-14

AI Summary

To address inefficient field-target search under battery constraints in agricultural drones, this paper proposes a lightweight deep reinforcement learning (DRL) path-planning method integrating uncertainty-aware prior knowledge. Based on the Proximal Policy Optimization (PPO) algorithm, a policy network is trained in an abstract simulation environment and deployed end-to-end on real hardware by coupling YOLO-based object detection with a ROS-based flight control system. Our key contribution is the first incorporation of uncertainty priors into an active-search RL framework for agriculture, augmented with a dynamic termination mechanism to accommodate low-recall-tolerance applications (e.g., weed monitoring). Experiments demonstrate that the method reduces traversal path length by 78% in simulation (with a 14% recall drop) and by 72% in real-field trials (with a 25% recall drop), validating both its effectiveness and practical deployability.

Technology Category

• Search and Optimization: Learning to Search
• Intelligent Robots: Motion and Path Planning
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Drones are promising for data collection in precision agriculture, however, they are limited by their battery capacity. Efficient path planners are therefore required. This paper presents a drone path planner trained using Reinforcement Learning (RL) on an abstract simulation that uses object detections and uncertain prior knowledge. The RL agent controls the flight direction and can terminate the flight. By using the agent in combination with the drone's flight controller and a detection network to process camera images, it is possible to evaluate the performance of the agent on real-world data. In simulation, the agent yielded on average a 78% shorter flight path compared to a full coverage planner, at the cost of a 14% lower recall. On real-world data, the agent showed a 72% shorter flight path compared to a full coverage planner, however, at the cost of a 25% lower recall. The lower performance on real-world data was attributed to the real-world object distribution and the lower accuracy of prior knowledge, and shows potential for improvement. Overall, we concluded that for applications where it is not crucial to find all objects, such as weed detection, the learned-based path planner is suitable and efficient.

Problem

• Develop drone path planner for efficient object search in agriculture
• Train planner using RL with simulation and uncertain prior knowledge
• Balance shorter flight paths against lower recall in real-world data

Innovation

• Reinforcement Learning trained drone path planner
• Combines flight controller and detection network
• Abstract simulation with uncertain prior knowledge