VARIABLE LEARNING RATE NEUROMORPHIC GUIDANCE CONTROLLER FOR AUTOMATED TRANSIT VEHICLES

This paper describes a guidance controller for automated transit vehicles which operates at high speeds. A feedforward neural network with a back propagation algorithm for learning forms the basis of the controller. A controller with a variable learning rate, whose value depends on the operating parameters of the vehicle, is then described. Real-time learning rates are used to compute empirical relationships. Simulation results show that the vehicle can recover from initial offsets and follow a track within few seconds for certain vehicle speeds.

• Supplemental Notes:
  • Publication Date: 1995 IEEE Service Center, Piscataway NJ
• Corporate Authors:

  Concordia University

  Department of Mechanical and Industrial Engineering
  1455 de Maisonneuve Boulevard, West
  Montreal, Quebec  Canada  H3G 1M8
• Authors:
  • Rajagopalan, R
  • Minano, D
• Conference:
  • IEEE International Symposium on Intelligent Control (10th : 1995 : Monterey, Calif.). Proceedings
• Publication Date: 1995

Language

• English

Media Info

• Pagination: p. 435-440

Subject/Index Terms

• TRT Terms: Artificial intelligence; Automated vehicle control; Public transit
• Uncontrolled Terms: Controllers
• Subject Areas: Data and Information Technology; Public Transportation;

Filing Info

• Accession Number: 00793397
• Record Type: Publication
• Source Agency: UC Berkeley Transportation Library
• Files: PATH
• Created Date: Jun 13 2000 12:00AM