Performance of a planar robotic manipulator with discrete trinary revolute joints

Karl A Kirsch, Purdue University

Abstract

Discrete joint manipulators are a potential alternative to common continuous joint robots used in industry today. Applications of a discrete joint manipulator include proximity positioning tasks, space exploration, camera placement, and automation systems. Proximity positioning tasks are tasks that require moving to regions of interest but not necessarily to a unique location inside the region. Sorting parts into baskets is an example of a proximity positioning task. Discrete robots are potentially useful for small scale robots because it is difficult to implement sensors and motors at a small scale level. The goal of this thesis is to design, build, analyze, and test a robotic manipulator with discrete trinary joints. A trinary joint is similar to a discrete binary joint, but a trinary joint has three discrete outputs instead of two. A manipulator with discrete joints is made up of simpler parts and has the advantage of performing repeatable tasks without the need for control feedback. Results show that a discrete robot can successfully be used for proximity positioning tasks. The performance depends on the location of discrete points relative to the desired points and also on the repeatability of the experimental manipulator. This thesis covers the design and experimental results of the manipulator. The robot manipulator links are designed to be modular and compact. Because links can be connected at −90, 0, or +90 degrees, the same links can be used to obtain different workspaces. After analyzing multiple manipulators with different sets of geometric parameters, the final discrete robot is a 7 link robot designed to work in an 8 x 12 inch gridspace. Results obtained via a vision system demonstrate the potential capabilities of the planar discrete manipulator. Results include subsets of end-effector positions in the gridspace and training results. Trajectories are also controlled by adjusting the timing sequence of actuation. Lastly, a computer model for simulating the trajectories using Lagrange Multipliers is developed and compared to experimental results.

Degree

M.S.M.E.

Advisors

Cipra, Purdue University.

Subject Area

Mechanical engineering|Robotics

COinS