Browsing by Author "Steed, Clint Alex"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Evaluation of vision-based robot control configurations for reconfigurable assembly
    (Stellenbosch : Stellenbosch University, 2016-03) Steed, Clint Alex; Basson, A. H.; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
    ENGLISH ABSTRACT: This thesis considers vision based robot control for a reconfigurable manufacturing cell. Reconfigurable manufacturing systems are aimed at rapidly adapting to fluctuations in market demand, shorter product life cycles and product customization. Computer vision is a promising component of such manufacturing systems, in particular aiding the flexibility to handle a range of products with reduced set-up time and fewer fixtures. The objective of the research presented here was to develop a simulation-based approach to compare eye-in-hand and fixed camera vision-based control within a reconfigurable assembly cell. These configurations have dissimilar system costs and system performance characteristics. The research used a KUKA six degree of freedom articulated arm robot, a DVT Legend 540 camera and a Cognex ism-1100 camera, as well as the software supplied with the cameras. The simulation was aimed at predicting the throughput rate of multiple eye-in-hand and fixed camera configurations, where configurations include varying the numbers of reconfigurable singulation units, positions of machines, parts being singulated, etc. The throughput rates for different configurations can then be compared to their costs and ease of reconfiguration. The simulation uses a holonic control system that was designed based on the ADACOR architecture. The thesis describes in detail the holons used, including their hardware, data structures and responsibilities. A multi-agent system was implemented for high-level control and some of the agents communicated with their respective lower-level controllers during validation testing. In the simulation, software (including KUKA Simpro and some custom software) replaced the hardware and low-level controllers. Tests on a number of physical laboratory configurations were used to validate the simulation. The simulation's application is demonstrated in a number of experiments in which cell configurations and machine performance were varied. For example, for a particular situation simulated, these experiments showed that the eye-in-hand configuration has a competitive cost to performance ratio when tasks have a significant waiting period (pallet exchange time, tasks deployed sparsely, etc.). In most experiments, however the fixed camera configuration performs better. The simulation allowed “hardware in the loop” testing, which also makes it a useful tool for development of the cell.