Real-time learning: a ball on a beam

作者: H. Benbrahim , J.S. Doleac , J.A. Franklin , O.G. Selfridge

DOI: 10.1109/IJCNN.1992.287219

关键词:

摘要: In the Real-Time Learning Laboratory at GTE Laboratories, machine learning algorithms are being implemented on hardware testbeds. A modified connectionist actor-critic system has been applied to a ball balancing task. The learns balance beam in less than 5 min and maintains balance. can roll along few inches of track flat metal beam, which an electric motor rotate. computer running PC senses position angular beam. prevent from reaching either end shown be robust through sensor noise mechanical changes; it also generated many interesting questions for future research. >

参考文章(2)
Andrew G. Barto, Richard S. Sutton, Charles W. Anderson, Neuronlike adaptive elements that can solve difficult learning control problems systems man and cybernetics. ,vol. 13, pp. 834- 846 ,(1983) , 10.1109/TSMC.1983.6313077
Charles W. Anderson, Learning to Control an Inverted Pendulum with Connectionist Networks american control conference. pp. 2294- 2298 ,(1988) , 10.23919/ACC.1988.4790107