The foregoing sections have sketched the rationale for and general design of an experimental training system based on automated adaptive principles for instructing student GCA controllers in the PAR phase of the final approach. The purpose of the laboratory facility will be to check out the functioning of the component subsystems and the overall training effectiveness of the total system. To this end it will be necessary to examine whether the technical solutions implemented fulfill the requirements that follow.

a. Speech Understanding Subsystem

1. Recognize speech inputs from the GCA phraseology automatically (in real time) with less than one per cent rejections.

2. Discriminate advisories which affect the aircraft simulation from those that do not.

3. Discriminate glidepath advisories from heading advisories.

4. Translate all advisories (with numerical values) into output terms acceptable to the Student Speech Evaluation Subsystem and those that affect the control of the aircraft into terms acceptable to the Training Control (Simulation) Subsystem.

b. Student Speech Evaluation Sybsystem

1. Generate the minimum number of performance measures required for prediction and diagnosis.

2. Generate a single score descriptive of performance on a run for the Adaptive Syllabus Control Subsystem.

c. Adaptive Syllabus Control Subsystem

1. Provide a syllabus exhibiting adequate problem coverage, difficulty range and difficulty resolution.

2. Provide an adaptive logic that maintains stable control and exhibits sensitivity to individual training needs.

d. Training Control (Simulation) Subsystem

1. Utilize the simplest (most quickly computed) set of equations that adequately represent the influences on aircraft control.

2. Generate the minimum visual presentation equivalent to the information content available to the student controller through the PAR display.

3. Provide for automatic initiation and termination of runs and sessions.

e. Total Experimental Facility

1. Provide means to assess effect of system on student motivation.

2. Provide capability to function in non-adaptive mode.

3. Provide capability to accept student input through means other than speech subsystem.

4. Provide means for "instructor" intervention.

The last four requirements, obviously, are intended to make possible studies to establish whether the system design implemented, beyond be able to "run", is able to train better than other designs lacking the full automated adaptive capabilities.

The purpose of this paper, in keeping with the spirit of the Industry Conference, has been to make you aware of our interest and activity in this area. Although we are hopeful that the Speech Understanding Subsystem will prove adequate we are particularly interested in any ideas you may have on speech understanding technology. Of course, your suggestions on any and all other parts of the project are most welcome.