An Approach Based on Particle Swarm Optimization for Inspection of Spacecraft Hulls by a Swarm of Miniaturized Robots

The remoteness and hazards that are inherent to the operating environments of space infrastructures promote their need for automated robotic inspection. In particular, micrometeoroid and orbital debris impact and structural fatigue are common sources of damage to spacecraft hulls. Vibration sensing has been used to detect structural damage in spacecraft hulls as well as in structural health monitoring practices in industry by deploying static sensors. In this paper, we propose using a swarm of miniaturized vibration-sensing mobile robots realizing a network of mobile sensors. We present a distributed inspection algorithm based on the bio-inspired particle swarm optimization and evolutionary algorithm niching techniques to deliver the task of enumeration and localization of an a priori unknown number of vibration sources on a simplified 2.5D spacecraft surface. Our algorithm is deployed on a swarm of simulated cm-scale wheeled robots. These are guided in their inspection task by sensing vibrations arising from failure points on the surface which are detected by on-board accelerometers. We study three performance metrics: (1) proximity of the localized sources to the ground truth locations, (2) time to localize each source, and (3) time to finish the inspection task given a 75% inspection coverage threshold. We find that our swarm is able to successfully localize the present sou