A Cockpit Natural Language Study - Selected Transcripts.
Abstract
This third report on the Cockpit Natural Language (CNL) study contains a brief description of the purpose and methodology of the CNL study, a section on lessons learned, scenario situation descriptions, a glossary and transcripts from 9 of the 54 pilots interviewed. Pilot comments cover the issues and implementation details of automation, displays, voice interaction and artificially-intelligent computer aids. Lessons learned from the CNL study include (1) Voice interaction is best employed as a new channel of information transfer, not just as a backup mode for manual or visual channels. For example, voice should be used at the intent-level (e.g., avoid SAM which means to re-plan route, altitude, etc. to avoid enemy surface-to-air missile detection), or to command multiple aircraft subsystems simultaneously (e.g. target helicopter, which means to achieve a radar lock on the helicopter and select an air-to-air missile), not just to replace a single switch actuation (e.g. select air-to-air missile), which is done quicker manually than verbally. (2) Pilot-cockpit voice interaction requires a shared information context between the pilot and the cockpit's computer in order to ensure that pilot commands are properly understood and executed. (3) Role-playing works (even with a low-fidelity cockpit simulation) when extracting valuable information from a pilot community. And (4), voice-activated computer messages should not replace pilot-to-pilot communications (radio calls, hand signals) because of the value of knowing that the other pilot (e.g., wingman) actually received the message.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1988
- Accession Number
- ADA192972
Entities
People
- Bryon T. Hollis
- Dan E. Flory
- David T. Williamson
- Michael P. Munger
- Ronald L. Small