UNCALIBRATED VISUAL SERVOING USING REAL-TIME VELOCITY OPTIMIZATION

摘要

A robotic control method for a camera (30) having an optical view and a robot (40) having an end-effector (42) and one or more joints (41) for maneuvering end-effector (42). The robotic control method involves an acquisition of a digital video frame (32) illustrating an image as optically viewed by the camera (30), and an execution of a visual servoing for controlling a pose of end-effector (42) relative to an image feature within the digital video frame (32). The visual servoing involves an identification of a tracking vector (vtrk) within an image coordinate system (80) of the digital video frame (32) extending from a tracking point (TR) to a target point (TG) associated with the image feature, a mapping of the tracking vector within a configuration space (100) constructed from a robotic coordinate system (90) associated with the end-effector (42), and a derivation of a pose of the end-effector (42) within the robotic coordinate system (90) from the mapping of the tracking vector (vtrk) within the configuration space (100).

主权项

1. A robotic system (20), comprising:
a camera (30) having an optical view for acquiring a sequence of digital video frames (32) illustrating an image; a robot (40) including an end-effector (42) and at least one robotic joint (41) for maneuvering the end-effector (42); and a robot controller (50) in communication with the camera (30) and the robot (40), the robot controller (50) including a visual servo (51) for controlling a pose of the end-effector (42) relative to an image feature within each digital video frame (32),
wherein, for each digital video frame (32), the visual servo (51) is operable to identify a tracking vector (vtrk) within an image coordinate system (80) of the digital video frame (32) extending from a tracking point (TR) to a target point (TG) associated with the image feature, to map the tracking vector (vtrk) within a configuration space (100) constructed from a robotic coordinate system (90) associated with the end-effector (42), and to derive a pose of the end-effector (42) within the robotic coordinate system (90) from the mapping of the tracking vector (vtrk) within the configuration space (100).