Adaptive Estimation of Information Values in Continuous Decision Making and Control of Remotely Piloted Vehicles
Abstract
This report describes research and development centered on evaluation of information needs in supervision of remotely piloted vehicles. The selection of information for transmission and display is a recurrent, subjective decision involving many factors - machine state, operator capabilities, communications costs, and channel limitations among others. An adaptive computer program has been developed which incorporates these factors into a multi-attribute decision model. The program is designed to capture the supervisory operator's decision policy by using a training algorithm based on pattern recognition techniques. Preliminary tests of the adaptive modeling approach were made using a task simulation resembling control of a remotely piloted vehicle. Individual subjects navigated the RPV through a changing, hazardous environment. In doing so, the operators selected different combinations of information and control allocation. The adaptive model was found to be more predictive of the subject's behavior than either a constant, unity weight model or an off-line method of weight estimation. Also, prediction of behavior increased with presentation of model- based recommendations to the subjects. Finally, the model was found to be useful in identifying differing decision strategies. The multi-attribute model thus formulated is expected to find application in evaluation of alternative information needs. Methods for management of communications by the remote element are also discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1977
- Accession Number
- ADA050414
Entities
People
- Amos Freedy
- Ken Chen
- Randall Steeb