Influence of path curvature on collision avoidance behaviour between two walkers

Area of Science:

• Human-computer interaction
• Robotics
• Cognitive psychology

Background:

• Navigating public spaces involves avoiding pedestrians with diverse movement patterns.
• Collision avoidance strategies for rectilinear paths are understood, but curvilinear paths remain under-explored.
• Predicting future interactions is crucial for safe navigation in crowded environments.

Purpose of the Study:

• To investigate collision avoidance behaviors when interacting with virtual humans following curvilinear trajectories.
• To compare avoidance strategies for curvilinear versus rectilinear paths.
• To determine if future distance of closest approach is a viable predictor for curvilinear interactions.

Main Methods:

• Twenty-two participants navigated a virtual environment, avoiding a virtual human (VH).
• The VH followed either rectilinear or curvilinear paths (5m or 10m radius curves).
• Collision occurrences and inversions in crossing order were recorded at varying closest approach distances.

Main Results:

• Curvilinear paths, especially with a 5m radius, resulted in more collisions (when VH approached from behind) and inversions (when VH approached from in-front) compared to rectilinear paths.
• Future distance of closest approach evolved similarly for rectilinear and 10m radius curvilinear paths.
• Participants demonstrated effective prediction of future closest approach distances for curvilinear trajectories.

Conclusions:

• Individuals can predict the future distance of closest approach for virtual walkers on curvilinear paths.
• Collision avoidance adaptations for curvilinear paths are comparable to those for rectilinear interactions.
• Findings suggest that current models of collision avoidance may be extended to include non-constant velocities and varied trajectories.