Gaze Behavior and Mental Workload While Using a Whole-Body Powered Exoskeleton: A Pilot Study

The mental demands associated with operating complex whole-body powered exoskeletons are poorly understood. Considering that powered exoskeletons may be intended for use in hazardous and complex industrial environments, where elevated mental workload may lead to distraction and reduced safety, it is important to better understand and quantify the potential cognitive demands required for control of these devices. Eye-tracking is a promising technique for measuring mental workload, since it can provide a variety of pupillary and gaze-behavioral metrics (Holmqvist et al., 2011) and is increasingly versatile and usable in the real world (Cognolato et al., 2018). Research on challenging coordination, movement-planning and postural-control tasks indicated that these tasks can influence cognitive demands and visuomotor strategies; specifically, pupil dilation (PD) increased (Kahya et al., 2018), the complexity of gaze (gaze entropy; GE) reduced (Walsh, 2021), and gaze was focused downward, towards the walking path (Koren et al., 2022) in order to improve stability. However, no research has quantified these effects in the context of using a whole-body powered exoskeleton (EXO).

Our study compared PD and gaze behavior while walking with an EXO on level ground (compared to without the EXO), as well as potential differences between first-time and skilled EXO operators. We hypothesized that PD and fixations on the path (PF) would be higher, and GE would be lower in the EXO condition compared to the no-EXO condition for first-time operators, and that the size of these effects would be attenuated for skilled operators.

A sample of 11 healthy male participants completed the study. Six participants were novices with no experience operating the whole-body powered exoskeleton prior to data collection. Five participants were experts who had extensive experience in testing and operating the whole-body powered exoskeleton throughout its developmental phases. We used an early prototype version of the Guardian^® XO^® (EXO for brevity) developed by Sarcos Robotics, which was specifically designed for heavy industrial applications. For measuring eye-activity, we used a head-worn Tobii eye-tracker (Tobii Pro glasses 2, Tobii Technology AB, Danderyd, Sweden) that sampled gaze data at 100 Hz.

Novices performed four level-walking sessions with the EXO and one without it (No-EXO session), with the first session involving familiarization and parameter tuning. Following this familiarization session, three subsequent walking sessions (S1-S3) were completed. In each walking session, data were collected when participants walked along a linear gait track at their preferred walking speed six times (i.e., six trials). Novices were randomly assigned to perform one no-EXO condition either in sessions 1 or 3. Experts completed two walking sessions (EXO and no-EXO) on the same day.

Verbal reports and observed walking performance indicated that walking with the exoskeleton was more mentally effortful than walking without. Considering eye-tracking measures, our hypothesis that novices would experience higher mental demand compared to experts, as reflected in PD and GE, was partially supported. Novices’ PD was significantly higher in the EXO session, whereas PD showed no difference in experts between the EXO and no-EXO sessions. This suggests that walking in the exoskeleton may have imposed greater motor-cognitive demands on novices than on experts. On the other hand, although the EXO condition led to a reduction in GE compared to the no-EXO condition, indicating less exploration of the visual environment due to greater cognitive demands (Walsh, 2021), this effect was similar in experts and novices. Thus, although the mental demands experienced by novices may be higher, the overall effects on visual exploration appear to be similar between the two groups.

We hypothesized that PF would increase while using the exoskeleton, indicating a greater visual focus on the walking path, possibly as a strategy to improve walking stability (Koren et al., 2022). As expected, PF increased in the EXO condition compared to the no-EXO condition for both novices and experts. However, interestingly, experts showed a much higher proportion of downward path-fixations compared to novices. It is possible that the use of downward-focused gaze as a strategy to improve walking stability may be learnt over time, and novices did not develop this strategy over the course of our study. Although downward-focused gaze can improve walking stability, it may also lead to reduced attention for monitoring the environment for potential hazards. In complex industrial environments, reduced attention towards the environment may put workers at greater risk of collisions with other workers and equipment. Thus, gaze behavior that serves to improve walking stability may, in fact, put the exoskeleton operator at a greater risk of collisions with surrounding objects. Future research should aim to quantify the actual impact of downward-focused gaze on safety, using methodologies such as peripheral detection tasks and dual-tasking while using powered exoskeletons.

To the best of our knowledge, this is the first study to investigate cognitive demands associated with walking in a whole-body powered exoskeleton, using eye-tracking measures. Our results also suggest that eye-tracking metrics could be used to identify expert strategies for using complex robotic devices. The ability to distinguish between expert and novice users, as well as changes in skill level, may promote the use of eye-tracking metrics as potential input variables for adaptive control strategies for powered exoskeletons.