Station-keeping using Perception and Relative Image-based Navigation and Tracking (SPRINT) for UAS

摘要

Under a recent DARPA project, SSCI performed initial design and testing of an innovative tightly-coupled vison and GNC system for follower vehicles to achieve safe approach and station-keeping with the lead vehicle within some range tolerance and inside a 60 degree cone, under leader maneuvers and vehicle capability constraints. The resulting system is referred to as the SPRINT (Station-keeping using Perception and Relative Image-based Navigation and Tracking), and is a fully autonomous system whose role is to safely and efficiently transition the follower vehicles from some known initial position to station-keeping, and to maintain the desired separation under lead vehicle maneuvers using vision only. The project specifically focuses on trade studies for performance analysis of the integrated system under different vision system properties and constraints, leader and follower vehicle capabilities, and characteristics of the proposed GNC algorithms under lead vehicle maneuvering. The trade studies included the effect of time delay, effect of target maneuvers, and the effect of camera parameters. Under the project, we have developed system architecture, requirements and metrics; vision-based algorithms for position and pose estimation; and guidance and control algorithms for safe approach and station keeping with the maneuvering leader. We also performed simulation analysis and testing at SSCI, demonstrating system-level feasibility of SPRINT to achieve project objectives. These accomplishments are described in the paper.