Development of an Intelligent Robotic Cell
In order for a robot to grip an object in a two dimensional environment, the gripping point and the rotation angle of a defined axis of the object are required. This paper presents the results of studies focused on the experimental requirements of an adaptive neural network learning technique for objects recognition and orientation detection. The results show that linking the neural network output to a robot controller achieves good system performance in picking up randomly placed objects. This project continues the current trend of developing intelligent machines in manufacturing.
Unable to display preview. Download preview PDF.
- 1.A.P. Reeves, R.J. Prokop, S.E. Adndrews and F.P. Kull (1988). The Dimensional Shape Analysis using Moments and Fourier Descriptor, IEEE Trans. Pattern Anal. Machine Intell. P937–943.Google Scholar
- 2.David Vernon (1991), Machine Vision-Automatic Vision Inspection and Robot Vision, Prentice Hall Published, p 140–154.Google Scholar
- 3.Patrick K. Simpson , Artificial Neural Network Systems, Pergramon Press, p 1.Google Scholar
- 4.Mclelland, J.L. and Rumellhart, D.E. (1988), Parallel Distribution Processing. The MIT Press, Vol 1. p. 444–460 and p. 318–364.Google Scholar
- 5.Mclelland, J.L. and Rumellhart, D.E. (1989), Explorations in Parallel Distributed Processing, MIT Press, p. 83–160.Google Scholar
- 6.C. Lee, P. Lisboa, K.O’Donovan , A Neural Network with Periodic and Probabilistic Coding for Component Orientation and Detection, 8th International Conference in System Engineering, Coventry, U.K.Google Scholar
- 7.C. Lee , A Neural Network for Control of Picking Up Random-Placed Objects, 1st International Conference on Automation Technology, Taipei, Taiwan, R.O.C. p 803–809.Google Scholar
- 8.Russell C. Eberhart, Roy W. Dobbins , Neural network PC Tools, Academic Press, p 35–49Google Scholar