A real-time Robot demonstration controlled by the BSP400 Neurocomputer
An actual implementation of a real-time, neural network controlled, robot car is presented in this paper. The simple car consists of two motors and 4 light sensors. Supervised learning behaviour of the car is achieved by using a neural network with adaptive connections. The car can be taught to avoid obstacles. The controlling neural network is implemented on the BSP400 neurocomputer, a Brain Style Processor with 400 nodes. A subset of the digital nodes in the BSP400 are connected by fixed weights to form logical circuits in order to re-train the car. In this way cooperative computation of both ’logical’ and ’neural’ processes are integrated into one system.
Unable to display preview. Download preview PDF.
- Heemskerk, J.N.H., J.M.J. Murre, J. Hoekstra, L.H.J.G. Kemna and P.T.W. Hudson, The BSP400: a modular neurocomputer assembled from 400 low-cost microprocessors. In: Artificial Neural Networks: Proceedings of the International Conference on Artificial Neural Networks (ICANN-91) Espoo Finland, T. Kohonen, K. Mäkisara, O. Simula and J. Kangas (Eds), Elsevier Science Publishers B.V. (North-Holland), 1991, 709–714.Google Scholar
- Heemskerk, J.N.H. & F.A. Keijzer. A real-time neural implementation of a schema driven toycar. (To be presented at the workshop on’ Schemas and Neural Networks: Integrating Symbolic and Subsymbolic Approaches to Cooperative Computation’, 19–20 October 1993, LA.)Google Scholar
- Murre, J.M.J. Learning and Categorization in Modular Neural Networks. Hemel Hempstad: Harvester Weatsheaf, and Hillsdale, NJ: Lawrence Erlbaum, 1992.Google Scholar