Algorithms for full 3-D robotic visual tracking of moving targets whose motion is 3-D and consists of translational and rotational components are presented. The objective of the system is to track selected features on moving objects and to place their projections on the image plane at desired positions by appropriate camera motion. The most important characteristics of the proposed algorithms are the use of a single camera mounted on the end-effector of a robotic manipulator (eye-in-hand configuration), and the fact that these algorithms do not require accurate knowledge of the relative distance of the target object from the camera frame. The detection of motion is based on a cross-correlation technique known as Sum-of-Squares Differences (SSD) algorithm. The camera model used introduces a number of parameters that are estimated on-line, further reducing the algorithms reliance on precise calibration of the system. An adaptive control algorithm compensates for modeling errors, tracking errors, and unavoidable computations delays which result from time-consuming image processing. Experimental results are presented to verify the efficacy of the proposed algorithms and to highlight the limitations of the approach. These experiments were performed using a multirobotic system consisting of Puma 560 manipulators.
Bibliographical noteFunding Information:
The ability to track moving objects using visual feedback is an important characteristic that many intelligent robotic systems must possess. This ability is often necessary if robotic systems are to inspect, grasp, or assemble objects in environments that are dynamically varying or inherently difficult to calibrate. Workspaces in which objects are transported on conveyer belts or in another moving robot’s grasp fall into this category, as well as underwater, toxic, and outer space environments. The main advantage that distinguishes visual sensors from other types of sensors and makes them particularly useful for these kinds of tasks, is that vision is a noncontact sensing mode that is able to provide information over a relatively large region of a system’s workspace. Even so, statically located visual sensors have a very limited working region when one considers the limited depth-of-field and spatial resolution that vision sensors typically possess. However, the working region of a visual sensor such as a CCD camera can be greatly extended if the camera is allowed to move while tracking and observing objects of interest. Simple 2- Manuscript received July 20, 1993; revised January 3, 1995. This work was supported by the Defense Advanced Research Rojects Agency under ARPA Order DAAA-21-89C-0001, by the National Science Foundation under Contract IRI-941OOO3 and Contract IRI-9502245, and by the Sandia National Laboratories under Contract AL-3021. The work of N. Papanikolopoulos was supported by the McKnight Land-Grant Professorship Award program. The work of B. Nelson was supported by a National Defense Science and Engineering Graduate Fellowship from the U.S. Army Research Office under Grant DAALO3-91-G-0272.