Self-customized electronic procedures for Just In Time training of space telerobotics

Abstract

Astronauts on Long Duration Missions (LDMs) will face complex problems for which they are untrained. Rehearsal may be unfeasible; the task need be completed on the first attempt and preparation is limited to a review of the electronic procedures (EPs). This motivates Just-In-Time Training (JITT): Astronauts learn- generic skills and EPs recombine these skills to train the new task immediately prior to execution. EPs typically have a fixed level of depth of detail, which ignores individual astronaut competence and the task's hierarchical step/sub-step structure. One astronaut may need details for all sub-steps, whereas another may simply refer to the highest level steps. By varying depth of detail, an astronaut may be able to customize the EPs to aid task performance by reducing extraneous cognitive load and focusing attention to salient features. The question is whether this approach reduces errors when a space telerobotics task is performed for the first time. To answer this, an experiment was carried out over two days on a desktop robotics simulator. On Day 1, all subjects (n=14) were trained to criterion on robotics skills, and were required to pass a screening assessment for continued participation in the experiment. On Day 2, JITT was given as a 30 minute period for procedure review before performing the task. Control group subjects were given non-alterable procedures, while the treatment group was able to customize. Customized JITT led to a lower error count (Mcontroi = 26.3, Mtreatment = 4.6, p = 0.023, mixed regression), and greater accuracy in adhering to the procedures (Mcontro = 82%, Mtreatment = 91%, p = 0.067, Welch's two-sample t-test, SDcontroi = 11%, SDtreatment = 3.6%, p = 0.014, f-test). Despite attempts to balance subject proficiency between groups, the treatment group was noted to exhibit a lower error rate during Day 1 training. So, while these results support the perspective that customization reduced extraneous cognitive load, there remains a potential confound of unbalanced groups. This experiment will help inform NASA training protocols for LDMs.