Technology Category
Intelligent Robots: Multimodal Perception & Sensor FusionSearch and Optimization: Learning to SearchComputer Vision: Motion & Tracking
Stone Soup Multi-Target Tracking Feature Extraction For Autonomous Search And Track In Deep Reinforcement Learning Environment
🤖 AI Summary
To address the challenge of coordinating heterogeneous sensor resources in future military airspace, this paper proposes an end-to-end sensor management framework based on deep reinforcement learning (DRL). The core innovation lies in the first-time configurable integration of the Stone Soup multi-object tracking (MOT) framework into a Gymnasium-based RL environment, serving as a high-fidelity, differentiable feature extractor that maps raw sensor measurements to track-state representations in real time. We train agents using PPO, SAC, and TD3 algorithms on joint search-and-track tasks, achieving significant performance gains over conventional heuristic policies. Experimental results demonstrate that this “tracking-feature-driven” DRL paradigm excels in policy effectiveness, generalizability across scenarios, and engineering deployability—providing a novel methodological foundation for autonomous airspace awareness.
Technology Category
Application Category
📝 Abstract
Management of sensing resources is a non-trivial problem for future military air assets with future systems deploying heterogeneous sensors to generate information of the battlespace. Machine learning techniques including deep reinforcement learning (DRL) have been identified as promising approaches, but require high-fidelity training environments and feature extractors to generate information for the agent. This paper presents a deep reinforcement learning training approach, utilising the Stone Soup tracking framework as a feature extractor to train an agent for a sensor management task. A general framework for embedding Stone Soup tracker components within a Gymnasium environment is presented, enabling fast and configurable tracker deployments for RL training using Stable Baselines3. The approach is demonstrated in a sensor management task where an agent is trained to search and track a region of airspace utilising track lists generated from Stone Soup trackers. A sample implementation using three neural network architectures in a search-and-track scenario demonstrates the approach and shows that RL agents can outperform simple sensor search and track policies when trained within the Gymnasium and Stone Soup environment.
Problem
Research questions and friction points this paper is trying to address.
Develops DRL training for sensor management in military air assets.
Integrates Stone Soup tracker with Gymnasium for RL training.
Demonstrates RL agents outperforming simple sensor policies.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Stone Soup tracking framework for feature extraction
Gymnasium environment for configurable RL training
Stable Baselines3 for efficient agent training
🔎 Similar Papers
No similar papers found.
