Multi-target search with incomplete information based on partial global path planning with Signal Caching And Rebound Exploration

Abstract

To address challenges such as global information loss, sparse target distribution, and environmental complexity, this paper proposes an efficient single-agent search strategy based on Signal Caching And Rebound Exploration (SCARE). The strategy enhances multi-target search efficiency by integrating a target signal information caching mechanism, a constrained motion pattern, and a path planning approach that synergizes local perception with global guidance. In signal-absent scenarios, the method employs the constrained walking space and confidence evaluation mechanism to redirect the robot and improve search coverage. Conversely, when signal conditions are available, a target-oriented strategy enhances target localization accuracy and efficiency. Extensive simulations, including ablation studies and comparative experiments, demonstrate the robustness and effectiveness of the proposed method. SCARE significantly outperforms baseline algorithms in diverse scenarios, achieving nearly 100% success rate with 4 targets in a 50 × 50 map containing 22% obstacles. Additional experiments validate the scalability to increasing target counts and obstacle densities, as well as its resilience against signal interference through enhanced caching mechanisms. These results highlight the method’s strong potential for deployment in complex and partially observable environments.