Skip to Main Content
This paper focuses on the development of a test platform for the capture of non-cooperative targets by an autonomous robotic manipulator. The paper describes the platform and subsystems involved in the capture operation with an emphasis on the vision system and controller. The vision system employs a custom algorithm, detecting the target, determining the quality of the lock and predicting the motion through the use of a Kalman filter. Photogrammetry uses the information obtained to determine the translational and rotational vectors linking the camera to the target that is fed into the controller that determines the optimal path for capture and deceleration. The controller is developed based on a hybrid force-impedance control that follows a force trajectory based on the motion of the target and decelerates the target smoothly while monitoring potential vibrations in the system. Currently, the system is being validated through a combination of simulations and experiments that take place on an air-bearing table as well as a mobile pendulum system that simulates free fall environments.