Sensor-based navigation of a mobile robot using automatically constructed fuzzy rules
A system for automatic generation of fuzzy rules is proposed which is based on a new approach, called ``Fuzzy Behaviorist,`` and on its associated formalism for rule base development in behavior-based robot control systems. The automated generator of fuzzy rules automatically constructs the set of rules and the associated membership functions that implement reasoning schemes that have been expressed in qualitative terms. The system also checks for completeness of the rule base and independence and/or redundancy of the rules to ensure that the requirements of the formalism are satisfied. Examples of the automatic generation of fuzzy rules for cases involving suppression and/or inhibition of fuzzy behaviors are given and discussed. Experimental results obtained with the automated fuzzy rule generator applied to the domain of sensor-based navigation in a priori unknown environments using one of our autonomous test-bed robots are then presented and discussed to illustrate the feasibility of large-scale automatic fuzzy rule generation using our proposed ``Fuzzy Behaviorist`` approach.
- Research Organization:
- Oak Ridge National Lab., TN (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC05-84OR21400
- OSTI ID:
- 10192726
- Report Number(s):
- CONF-9311107-2; ON: DE94001137
- Resource Relation:
- Conference: ICAR `93: 6. international conference on advanced robotics,Tokyo (Japan),1-2 Nov 1993; Other Information: PBD: [1993]
- Country of Publication:
- United States
- Language:
- English
Similar Records
A fuzzy behaviorist approach to sensor-based robot control
Adding memory processing behaviors to the fuzzy behaviorist-based navigation of mobile robots