Application to Animating a Digital Actor on Flat Terrain Joint variables are avars for primary motions. A Motion clip is a curve parameterized by time in the joint space (C-space). MPEG4-SNHC standard body description each hand = 25 DOF lower body = 21 upper body = 62 global hip position & orientation = 6 The actor The environment

Principle: Bound the actor by a cylinder and project all objects the ground Joint variables are avars for primary motions. A Motion clip is a curve parameterized by time in the joint space (C-space). MPEG4-SNHC standard body description each hand = 25 DOF lower body = 21 upper body = 62 global hip position & orientation = 6

Shrink the disc representing the actor to a point and grow the obstacles accordingly

How would you adapt the visibility graph method?

Motion Capture

Integration of Motion Planning and Motion Capture High-Level Navigation Goals Fast 2D Path Planner Path Consider the problem of controlling an animated character instructed to navigate towards a goal location on a level-terrain cluttered with obstacles Path-Following Controller Body Posture Graphic Display Base Point PD Controller Motion Capture Data

Extensions? Small obstacle Soft, but uniform terrain Gap in terrain Stairs ... Rough, irregular, possibly steep terrain

Simulated Vision Actor’s view KEY CONCEPT: Realistically model the information flow from the environment to the character. Complete access to all objects is both conceptually unrealistic and can be impractical for large environments Tasks involving navigation and obstacle avoidance require visual feedback Problem: Need to compute what objects are currently visible to an animated agent (3D visibility query). PREVIOUS WORK: Spherical sensors (C. Reynolds, ‘87) Ray-casting techniques (Tu & Terzopoulos ‘94/ C. Reynolds ‘91/ Funge, et al. ‘97) False-color Rendering and dynamic octrees (Noser, et al. ‘92) Iterative pattern-matching and color histograms (Terzopoulos & Rabie ‘96) Image-based motion energy (B. Blumberg ‘96) Actor’s view

Perception-Based Planning GOAL: Allow a character to explore an unknown interactive virtual environment. Character should maintain a visual memory or “cognitive map” of what it has perceived through the synthetic vision module. This map can subsequently be used as input to a navigation planner. The visual memory data structure must be simple and efficient to access and update. The navigation plan is updated as the character explores the environment Fast Path Planner Path-Following Controller Base Point PD Controller Motion Capture Data Simulated Vision Obstacles

Perception-Based Planning GOAL: Allow a character to explore an unknown interactive virtual environment. Character should maintain a visual memory or “cognitive map” of what it has perceived through the synthetic vision module. This map can subsequently be used as input to a navigation planner. The visual memory data structure must be simple and efficient to access and update. The navigation plan is updated as the character explores the environment Plan Sense Act

Actor that remembers and learns Should a digital actor remember the objects it sees and their locations? How could it detect that the environment has changed between two visits? What could it learn about objects?