Design and analysis of cooperating mobile robots using non-grasping contact for object transportation

Tyson L Ringold, Purdue University

Abstract

Cooperative robotics is concerned with the simultaneous utilization of multiple robots and promises a number of advantages, including increased robustness, decreased complexity of the robots themselves, and, through synergistic effects, better execution of a task over a single, similarly capable robot. The task of object transportation is especially well suited to cooperative robots since large, bulky items that require correspondingly large handling devices can instead be supported by several significantly smaller robots. Due to their inherently limited resources, mobile robots benefit the most from cooperation. For the task of object transportation, a unique type of mobile robot is presented which, when part of a group, is able to lift and carry objects that may be considerably larger than itself. A key feature of the presented robot is that it is devoid of articulated mechanisms, and instead interacts with objects through the non-grasping, frictional contact of wheels. This ultimately allows the robot to transport objects without any special features or fixturing points, and by way of relative motion, allows for unique motions that permit traversal across cluttered, unstructured terrain. To determine when such a robot can lift an object, a static force analysis is initially performed. Through the use of contour plots, maps of configurations where the process is feasible are generated, allowing quick visual inspection of the robot's ability to lift objects. In order to quantitatively demonstrate that a group of these robots are physically capable of controlling an object, a new formulation for robotic grasp analysis is established as well. This technique can be used to find the best set of joint torques that optimizes an arbitrary metric, even when the robots are kinematically deficient and redundancies exist. Beyond the team of robots addressed in this work, the technique is generalizable to any multi-contact problem, such as robotic hands, fixtures, and whole-arm grasps. To control the team of the proposed robots, a behavior-based control scheme is devised for a lifting maneuver and an extension of the artificial potential field method is developed for cooperative carrying of an object. These control strategies do not rely on a single controller, but instead allow each robot of the team to determine its motions independently, making the team scalable to an arbitrary number of robots. The strategies are then tested through a number of dynamic simulations, showing that the process is robust to errors and benefits from the use of many robots.

Degree

Ph.D.

Advisors

Cipra, Purdue University.

Subject Area

Mechanical engineering

COinS