Planning Motions and Placements for Virtual Demonstrators.

Area of Science:

• Robotics
• Computer Graphics
• Human-Computer Interaction

Background:

• Virtual human demonstrators require sophisticated motion planning to navigate complex environments and mimic human coordination.
• Existing systems often struggle with integrating locomotion, action, and gaze while respecting spatial constraints.

Purpose of the Study:

• To develop a whole-body motion planning and synthesis framework for virtual humans performing demonstrations.
• To achieve human-like coordination and address spatial constraints in interactive applications.

Main Methods:

• A coordination model derived from human subject experiments was used.
• The planner identifies body placements considering visibility, locomotion, and action feasibility among obstacles.
• Actions are modeled using motion clusters, with a collision avoidance procedure in blending space.

Main Results:

• The framework successfully coordinates locomotion, body positioning, action execution, and gaze for demonstration tasks.
• Human-like solutions were achieved by integrating a human-derived coordination model.
• The system effectively handles obstacle-rich environments and ensures visibility constraints.

Conclusions:

• The proposed framework offers effective approaches for modeling and solving complex demonstrative tasks for virtual humans.
• This work advances the capability of virtual agents to perform realistic and interactive demonstrations.