Pose Estimation for Autonomous In-Space Assembly

The expansion of humanity's exploration necessitates new methods of building and deploying space structures. The future of in-space servicing, assembly, and manufacturing requires the contribution of autonomous systems with the capability to perceive their environment to accomplish complex tasks. The work presented here consists of a novel method for processing pose estimate data from ArUco markers using a temporally-averaged fusion approach. This algorithm has been implemented on the Precision Assembled Space Structures project for estimating the pose of a tri-truss during a manipulator fine alignment step. Preliminary results compare the pose difference between the camera and a 3D marker board with a ground truth pose difference from known dimensions of a testing rig. The results indicate that the data fusion method is successful in mapping a set of measurements across multiple markers into a singular pose estimate for a given tri-truss. The results also indicate that future work should focus on improving the accuracy of individual camera-to-marker transformation estimates to further reduce the error in the overall pose estimate.