用微型振动机器人集群检测金属表面损伤，靠振动模式差异定位问题。

Robot swarms offer the potential to serve a variety of distributed sensing applications. An interesting real-world application that stands to benefit significantly from deployment of swarms is structural monitoring, where traditional sensor networks face challenges in structural coverage due to their static nature. This paper investigates the deployment of a swarm of miniaturized vibration sensing robots to inspect and localize structural damages on a surface section within a high-fidelity simulation environment. In particular, we consider a 1 m x 1 m x 3 mm steel surface section and utilize finite element analysis using Abaqus to obtain realistic structural vibration data. The resulting vibration data is imported into the physics-based robotic simulator Webots, where we simulate the dynamics of our surface inspecting robot swarm. We employ (i) Gaussian process estimators to guide the robots' exploration as they collect vibration samples across the surface and (ii) operational modal analysis to detect structural damages by estimating and comparing existing and intact structural vibration patterns. We analyze the influence of exploration radii on estimation uncertainty and assess the effectiveness of our method across 10 randomized scenarios, where the number, locations, surface area, and depth of structural damages vary. Our simulation studies validate the efficacy of our miniaturized robot swarm for vibration-based structural inspection.