Deutsch
 
Hilfe Datenschutzhinweis Impressum
  DetailsucheBrowse

Datensatz

DATENSATZ AKTIONENEXPORT

Freigegeben

Meeting Abstract

Restricted field of view during training impacts gaze strategy for aircraft handling

MPG-Autoren
/persons/resource/persons83861

Chuang,  LL
Project group: Cognition & Control in Human-Machine Systems, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;
Department Human Perception, Cognition and Action, Max Planck Institute for Biological Cybernetics, Max Planck Society;

Volltexte (beschränkter Zugriff)
Für Ihren IP-Bereich sind aktuell keine Volltexte freigegeben.
Volltexte (frei zugänglich)
Es sind keine frei zugänglichen Volltexte in PuRe verfügbar
Ergänzendes Material (frei zugänglich)
Es sind keine frei zugänglichen Ergänzenden Materialien verfügbar
Zitation

Walter, J., & Chuang, L. (2018). Restricted field of view during training impacts gaze strategy for aircraft handling. Frontiers in Human Neuroscience, 2018(Conference Abstract: 2nd International Neuroergonomics Conference).


Zitierlink: https://hdl.handle.net/21.11116/0000-0001-7E3B-1
Zusammenfassung
Vehicle simulation is an important instrument for the acquisition and maintenance of complex skills (e.g., aircraft handling; Salas, Bowers & Rhodenizer, 1998). Such training not only improves control performance (i.e., error- or time-based measures), but also alters gaze strategies (Gegenfurtner, Lehtinen & Säljö, 2011). It is claimed that eye movements can reflect the strategic allocation of attention (Hornof & Halverson, 2003). For example, experts exhibit more consistent gaze pattern than novices (Kasarskis et al., 2001). More importantly, flight training can reduce initial differences in gaze patterns between expert and novice pilots (Harris, Glover & Spady, 1986). Thus, training modulates gaze strategies such that eye movements get more appropriate for the goals of vehicle handling. Unfortunately, training simulations often deviate from the real world, such as available field-of-view (FOV) (Moroney & Lilienthal, 2008). FOV conditions can influence eye-head coordination during visuo-manual tracking tasks (Sandor & Leger, 1991). Hence, we ask: Do gaze strategies acquired under reduced FOV conditions, as it might be the case in simulator settings, persist when FOV restrictions are removed? To answer this question, we recorded the eye movements of participants (N=24) who performed a lateral translational movement in 3D space with a simplified rotorcraft, using a head-mounted eye-tracking system (SMI ETG, Sensorimotoric Instruments GmbH; sampling rate: 30 Hz). The experiment took place in a front-projection CAVE environment which enables large FOV. The environment consisted of an airfield with a lateral arrangement of vertical poles that indicated the ideal altitude as well the operator’s desired path trajectory (see Figure 1). Moreover, points on the ground explicitly indicated the desired path between the start- and end-zone. In a between-subject design, three participant groups received 30 training trials under three different levels of FOV restriction. Subsequently, they were equally tested over ten trials with a full FOV condition (230° x 125° visual angle). Critically, training conditions differed in terms of the amount of information that was available during flight. Participants who were trained under FOVsmall (60° x 60° visual angle) could only look two rods ahead without being able to see the ground. FOVvert (180° x 60° visual angle) allowed the operators to look ahead with regards to the vertical markers but not to see the ground path. FOVlarge (180° x 180° visual angle) allowed the operators to look ahead with regards to the vertical and ground markers. Leftwards and rightwards lateral movements were performed equally often. To compare the effects of FOV restriction on gaze strategies, only the final FOVfull condition was analyzed. If eye movements are generic, groups should not differ in gaze strategies under FOVfull as gaze pattern should not depend on FOV condition during training. However, gaze patterns during testing should differ if training FOVs have an influence on gaze strategies that persist even with the availability of a larger FOV. We defined four region of interests (ROI) that comprised the areas spanned by the poles that are visible in the FOVsmall training condition (poles near), the area between poles that lie outside of the FOVsmall (poles far), an area spanned by the ground markers which are less than two poles away (ground near) as well as the area between ground markers which are more than two poles away (ground far). We employed linear mixed-effects models (Krueger & Tian, 2004) for each ROI to test for differences between FOV conditions during training in the number of dwells and the average dwell duration, respectively. For each ROI, we formulated multiple models using RStudio. The full model contained fixed effects for trials (i.e. time) and FOV training condition, an interaction effect between trials and FOV training condition, and random slopes and intercepts for participants and flight direction. Two more parsimonious models differed from the full model only with respect to the interaction effect (model 1) and the interaction effect as well as the fixed effect for FOV training condition (model 2). We conducted Likelihood-Ratio tests (full model vs. model 1 and model 1 vs. model 2) for each ROI to test for significance of the interaction effect and the main effect of FOV training condition, respectively. Figure 2 shows that the FOV condition during training influenced gaze pattern during testing (FOVfull). Reliance on visual cues was influenced by available FOV size during training, whereby smaller FOV sizes resulted in a stronger reliance on proximal cues (near poles and near ground) and larger FOV sizes promoted the use of distant cues (far poles and far ground; see Figure 2a). Participants who received training under FOVsmall condition spent a significantly larger amount of dwells on proximal poles than the FOVlarge group did (χ2(2) = 8.82, p < .05). The trend in proportion of dwells on round near reflected the same tendency but was not significant (χ2(2) = 2.26, p = .32). In contrast, larger FOVs during training resulted in participants who relied on distant cues. However, this tendency was not significant for far poles (χ2(2) = 0.37, p = .83) or far ground (χ2(2) = 1.85, p = .40). There was no significant interaction effect for any of the ROIs (all p > .1). Thus, participants did not vary their gaze strategies during the test trials. Next, we analyzed the effect of FOV size during training on dwell duration. Results for dwell time showed a significant difference in the dwell time on near poles (χ2(2) =12.12, p < .01), with FOVlarge leading to a significantly lower dwell time on near poles (see Figure 2b). There were no other significant effects of FOV during training nor any significant interaction effect (all p > .1). To summarize, FOV conditions shape gaze patterns that stay stable even when FOV conditions change. These findings underscore the relevance of naturalistic simulation conditions during training, especially in terms of FOV. As eye tracking has been shown to be a sensitive marker of pilot’s situation awareness and attention allocation (Sarter, Mumaw & Wickens, 2007; Hanson, 2004), these new insights might have important implications for both, the design of training devices and the assessment of training efficiency.