Modeling Locomotion with Body Angular Movements in Virtual Reality

Research Background and Issues

• Identified Problems or Challenges: This paper highlights the lack of research on modeling completion time for virtual displacement tasks in virtual reality (VR) environments. While many interaction tasks such as pointing, crossing, and turning already have time prediction models, modeling navigation time in VR remains an unexplored area. At the same time, direction control and speed control are core factors in virtual positioning, yet there is a lack of effective integrated models.
• Significance: Accurate time prediction models can help optimize VR design, such as evaluating difficulty or providing improvement suggestions based on completion time during navigation. This has significant implications for the development of VR games, simulation training, and virtual navigation tools.
• Research Motivation and Related Work: The authors focus on VR navigation methods assisted by body-oriented motion, as this approach enhances users' sense of direction and presence while reducing motion discomfort and task load compared to other methods (e.g., teleportation or walking based on target point selection). Theoretical work on multi-stage path navigation and turning time is still underdeveloped in the field, motivating the authors to address this research gap.

Solution

• Proposed Method or Solution: This paper proposes a time prediction model to estimate the completion time for users navigating multi-segment zigzag paths in virtual environments. The model is divided into two main components: navigation along multi-segment paths and turning at path intersections.
  • The first component is divided into three stages (acceleration, maximum speed, deceleration), with each stage's speed-distance relationship defined by an exponential model from kinematics.
  • The second component reveals a linear relationship between turning angle and completion time.
  • The model structure quantifies the overall time through a comprehensive formula combining the two components.
• Innovations:
  • An exponential model is introduced in the acceleration and deceleration stages to more precisely fit the speed-distance curve.
  • A unified time calculation formula is proposed, integrating task variables such as speed, path length, and turning angle into a single framework.
  • By segmenting motion stages and incorporating turning angles into a comprehensive model, it is more applicable to real-world scenarios compared to existing models (e.g., TTA and PL models).
• Implementation Steps and Key Techniques:
  • Analyze the speed-distance curves for acceleration, maximum speed, and deceleration stages, and validate the goodness of fit of the exponential model through regression analysis.
  • Separate path navigation and turning tasks, establishing prediction formulas for estimating the time of each part.
  • Conduct three experiments to validate the model's effectiveness, including data analysis under different path lengths, angles, and linear transformation coefficients.

Research Outcomes

• Specific Results:
  • Experimental results show that the proposed model significantly outperforms baseline models (TTA and PL models) in time prediction performance.
  • For single-segment navigation time prediction, the model achieves a high goodness of fit (𝑅² values exceeding 0.987) with minimal deviation from actual times.
  • The linear relationship model between turning time and angle performs well, with a goodness of fit exceeding 0.989.
  • Validation results of the comprehensive model indicate a high correlation between predicted and actual times in multi-segment path navigation (𝑅² values exceeding 0.914), with a root mean square error of 3.282 seconds for prediction errors.
• Advantages Compared to Existing Solutions:
  • The proposed model integrates the relationship between path navigation and turning, comprehensively covering the two core subtasks of virtual positioning.
  • The new model distinguishes between acceleration and deceleration stages, significantly improving prediction accuracy, particularly in short paths and complex turning scenarios.
  • Unlike the TTA model, which only handles fixed-speed scenarios, this model adapts to changes in motion speed and task complexity.
• Limitations and Future Directions:
  • The model does not account for the impact of obstacles or uneven surfaces in complex scenarios, which may introduce additional navigation delays.
  • The model primarily targets tasks with defined paths and needs further extension for curved path tasks.
  • Current experiments are limited to young participants, and the model's applicability to different age groups (e.g., older adults) has yet to be verified.
  • The model design does not address cognitive delays (e.g., time losses due to distractions), and future research should incorporate cognitive-related variables.

Conclusion

This study holds significant importance in the current field of virtual reality, filling the gap in time prediction models for virtual navigation tasks and providing precise tools and practical guidance for related applications (e.g., educational training, navigation system optimization). Further research on adapting the model to complex environments and specific populations will be a key focus for future work.