Effects of Peripheral Visual Cues in Simulator-Based Training of Multimodal Control Skills

Abstract

This paper describes a training experiment performed in the SIMONA Research Simulator at Delft University of Technology to evaluate the effectiveness of peripheral visual cues as a substitute for physical motion feedback during the development of multimodal control skills. Twenty task-naive participants were divided in two experimental groups and performed a skill-based compensatory roll tracking task. Both groups were trained in a fixed-base setting, but one group was provided with additional out-of-the-window peripheral visual (roll rate) cues. After training, participants were transferred to a moving-base condition where pure roll motion cues were provided. The development of skills in both groups throughout the experiment was studied and compared using fits of a multimodal human operator model with an error and a roll feedback response, to explicitly quantify operators' use of all supplied feedbacks. As expected, the group that had access to peripheral visual cues attained better tracking performance during training and showed the development of multimodal control behavior. However, no evident transfer of this developed multimodal control strategy to a motion setting was observed. This suggests that training with peripheral visual cues is not an effective substitute for training the multimodal skills human operators use when motion feedback is available.

Effects of Motion Cues on the Training of Multi-Axis Manual Control Skills

“…18,24,25 Pilot control activity for roll and pitch are shown in Figures 7a and 7b was not expected based on previous training studies, for which control activity increased when training with motion. 12 The RMS of the roll control input during training was higher for HO and control activity at the end of training was more similar to that after transfer. The group training with HB motion showed an increase in roll control intensity after the training phase.…”

“…This indicates more runs might have been required to reach asymptotic levels for some of the parameters characterizing manual control behavior, as was also found in previous quasi-transfer-of-training experiments using a cybernetic approach. 11,12 In addition, even though pilots were assigned to one of the two training-motion conditions randomly, this might point to other differences between the groups, like differences in experience. The much lower pitch control VAF for motion group HO in both the training and transfer phases, indicating this group of pilots possibly controlled less linearly, might be an indication of this.…”

“…4,6,23 Similar effects of motion were also observed in quasi-transfer-of-training experiments using single-axis manual control tasks. 11,12 Based on these previous observations, the following hypotheses were formulated for the current experiment:…”

“…Only few studies have used this approach in the context of training and the ones that did used single-axis control tasks, 9,10 or looked at the effects of elementary variations of visual and motion cues, such as the effects of training with and without motion. 11,12 Even though these studies are crucial for our understanding of the acquisition of manual control skills, their results might not directly apply to the training and transfer of training of manual control skills for aircraft flying tasks that mostly require control and result in motion cues in multiple axes simultaneously.…”

Retention of Manual Control Skills in Multi-Axis Tracking Tasks

“…The considered exponential learning curve model is given by Eq. 5and has successfully been applied in earlier training studies to quantify learning trends [31,41,46,57].…”