AAC: An Acoustic Actor-Critic Trajectory Planning and Correction System for Stable Multi-Particle Levitation Displays

要旨

Acoustic levitation enables mid-air displays using physical particles to create 3D visuals, but stability limits the achievable animation complexity. Stability depends on factors including the acoustic solver, particle count, motion speed, and path geometry. This paper analyzes these factors, characterizing their effects, identifying constraints, and allowing particles to successfully follow the paths. We then propose Acoustic Actor-Critic (AAC), a closed-loop motion planning system that maximizes stability for multi-particle trajectories with minimal changes to the intended visual content. This follows a plan-detect-repair strategy: i) the Actor plans trajectories under the established constraints; ii) the Critic evaluates their stability and detects instabilities; iii) the Repair modules trigger localized repairs upon unstable path segments. Results showed that AAC can automatically refine and repair multi-particle trajectories, reducing failures from 21\% to 6\% across 100 paths. Our findings enable creators to produce more stable levitation paths, while AAC automatically refines trajectories with minimal deviation from the original animations.

著者
jincheng wang
UCL, LONDON, United Kingdom
Lei Gao
University College London, London, United Kingdom
Zhouyang Shen
University college london, London, United Kingdom
Pengyuan Wei
University College London, London, Select your state/province/region, United Kingdom
Diego Martinez Plasencia
University College of London, London, United Kingdom

会議: CHI 2026

ACM CHI Conference on Human Factors in Computing Systems

セッション: Shape-Changing Systems

P1 - Room 124
7 件の発表
2026-04-16 18:00:00
2026-04-16 19:30:00