Methods for Evaluation of Human-in-the-Loop Inspection of a Space Station Mockup Using a Quadcopter

Suited astronauts currently perform visual inspections for external spacecraft damage during extravehicular activity (EVA). Small, free-flying satellites can be used to perform these visual inspection tasks to reduce crew risk by keeping astronauts inside the vehicle. One approach gives the astronaut teleoperational control of the satellite, while another approach is engaging a control system that commands the satellite to fly along an inspection path. We assembled a mockup of a space station (featuring three modules of varying geometry) and used a quadcopter fitted with a GoPro Camera to emulate visual inspection of the mockup by the small satellite. A pilot participant performed the space station inspection task by utilizing the teleoperation mode of the quadcopter. The participant had access to a line-of-sight view of the quadcopter and space station mockup from a distance, and an egocentric view of the testing area (including the mockup) from the quadcopter's onboard camera. The inspection task required the participant to actively scan for, locate, and take pictures of surface anomalies on the space station mockup. Performance was evaluated by determining percentage of anomalies detected, time taken to complete the inspection, and gaze time percentages (physical quadcopter and space station versus camera display). Observations were recorded for the teleoperated flight mode and indicate a strong user preference for an egocentric quadcopter view to support anomaly detection and navigation. In the teleoperated flight mode, the participant had difficulty with accurately and consistently locating anomalies while also navigating and controlling the quadcopter. Global situation awareness abilities and anomaly detection performance were limited, suggesting further research is needed to support human-robot teaming for spacecraft inspection. This preliminary evaluation supports continued data collection with the developed test bed for human-robot interaction with additional levels of automation. Findings from the methods proposed can lead to design recommendations regarding the usage of teleoperated and autonomous flight modes as well as the impact of available viewpoints.