Fig 1.
Experimental setup and calibration.
(A) 10 right-handed healthy subjects performed 300 repeated trials of billiards shots in Embodied Virtual Reality (EVR). Green arrows mark the motion capture markers used to track and stream the cue stick movement into the EVR environment (B) Scene view in the EVR. Subjects were instructed to hit the cue ball (white), which was a physical ball on the table (in A), in attempt to shoot the virtual target ball (red) towards the far-left corner. (C) For environments calibration, MoCap markers were attached to the HTC Vive controllers which were placed in the pool-table’s pockets with additional solo marker in the cue ball position.
Fig 2.
With-in shot comparison between real cue ball trajectory (measured with a high-speed camera) and the virtual ball trajectory (calculated by the VR physics engine). Each one is a shot. A total of 100 billiard shots at various directions (-50˚< ø < 50˚ when 0 is straight forward) are presented. (A) Cue ball angles. (B) Max velocity of the cue ball during each trial. The regression line is in black with its 95% CI in doted lines. Identity line is in light gray.
Fig 3.
Task performance in EVR vs real-world.
(A) The mean absolute directional error of the target-ball, (B) The success rate, (C) directional variability, and (D) directional variability corrected for learning (see text). (A-D) presented over blocks of 25 trials. Solid lines present the performance of 10 subjects in the novel EVR environment. For compression, dashed lines present the performance of a group of 30 subjects in the same pool paradigm in the real-world (with no VR) from our previous study [21].
Fig 4.
Velocity profiles in EVR vs real-world.
Velocity profiles in 3 degrees of freedom (DoF) for each joint of the right arm joints. Blue lines are the profiles during the first block (trials 1–25), and red lines are the velocity profiles after learning plateaus, during the ninth block (trials 201–225). Solid lines present the velocity profiles of 10 subjects in the novel EVR environment. For compression, dashed lines present the velocity profiles of a group of 30 subjects in the same pool paradigm in the real-world (with no VR) from our previous study [21].
Fig 5.
Variance and complexity comparison.
(A) The variance covariance matrix of the right arm joints velocity profiles in EVR, averaged across subjects and trials over the first block, second block and the ninth block (after learning plateaus). (B) The trial-by-trial generalized variance (GV), with a double-exponential fit (red curve). (C) The number of principal components (PCs) that explain more than 1% of the variance in the velocity profiles of all joints in a single trial, with an exponential fit (red curve). (D) The manipulative complexity (Belić and Faisal, 2015), with an exponential fit (red curve). (B-D) Averaged across all subjects over all trials. Data is averaged over the performance of 10 subjects in the novel EVR environment. Grey dots are the trial averages for the EVR data. Solid red lines are curve fits for EVR data. For compression, dashed lines present the curve fits to a group of 30 subjects in the same pool paradigm in the real-world (with no VR) from our previous study [21].
Fig 6.
Velocity Profile Error (VPE) and Intertrial variability reduction across all joints in the EVR task. (A) The trial-by-trial VPE for all 3 DoF of all joints, averaged across all subjects, with an exponential fit. The time constants of the fits are reported under the title. Color coded for the DoF—blue: flexion/extension; red: abduction/adduction; green: internal/external rotation. (B) VPE intertrial variability (ITV) over blocks of 25 trials, averaged across all subjects. Data is averaged over the performance of 10 subjects in the novel EVR environment.