AcoustoBots: A swarm of robots for acoustophoretic multimodal interactions

Technical Summary: AcoustoBots – A Swarm of Robots for Acoustophoretic Multimodal Interactions

Problem Statement
Acoustophoresis, the use of sound waves to manipulate objects and generate haptic, audio, and visual effects in mid-air, has traditionally been implemented using singular, static phased array of transducers (PAT). While effective for specific tasks like levitation, volumetric displays, and mid-air haptics, this static implementation suffers from significant limitations regarding scalability, mobility, and versatility. Commercial systems are often confined to fixed, small interaction volumes (e.g., 63 × 48 × 48 cm), restricting dynamic interactions in expansive environments. Furthermore, static configurations hinder the ability to adaptively target multiple users, hand over levitated content between surfaces, or support complex multimodal scenarios where interaction zones must shift dynamically. The authors identify a need for a mobile, scalable, and modular acoustophoretic system that can overcome these spatial and functional constraints.

Methodology
To address these limitations, the authors introduce "AcoustoBots," a swarm robotic platform that integrates acoustophoretic capabilities with mobile robotics. The system architecture consists of the following key components:

• Hardware Design: Each AcoustoBot is a self-actuated, battery-powered robot based on the open-source "Mona" robot chassis. Mounted on the robot is a custom 8x8 mini-PAT board containing 40 kHz ultrasonic transducers. The board is controlled by an Altera Cyclone IV FPGA for precise phase and amplitude modulation, managed by an ESP32 microcontroller for wireless communication.
• Dynamic Actuation: A novel hinge actuation system, driven by a stepper motor, allows the mini-PAT board to rotate. This enables the array to switch between three orientations: horizontal (0°) for haptic interactions, sloped (45°) for audio interactions, and vertical (90°) for acoustic levitation.
• BeadDispenserBot: To automate the levitation process, the authors designed a specialized "BeadDispenserBot." This robot carries a bead dispenser mechanism (using a stepper motor and carousel) to deliver particles (e.g., expanded polystyrene beads) directly into the acoustic traps generated by the AcoustoBots.
• Control and Communication: The system operates on a client-server architecture. A central Server PC runs the swarm control algorithms, acoustic field generation (using a piston model), and tracking logic. It communicates via UDP over a WiFi network with the distributed clients (AcoustoBots, BeadDispenserBot, and a PhaseSpace tracking system). The tracking system monitors the real-time position and orientation of all bots and user hands/ears to facilitate dynamic alignment and following behaviors.
• Experimental Setup: The authors evaluated the system using two AcoustoBots and one BeadDispenserBot in a controlled environment. They tested three distinct scenarios: independent haptic feedback, collaborative directional audio, and cooperative acoustic levitation.

Key Contributions

1. AcoustoBot Platform: The design and development of a swarm robotic platform that combines mobility with acoustophoretic functionalities, enabling multimodal interactions (haptics, audio, levitation) in expanded, dynamic spaces.
2. Hinge Actuation System: A novel mechanical design allowing the mounted PAT boards to dynamically change orientation, facilitating the interchange between different interaction modalities (haptics, audio, levitation) within a single swarm.
3. BeadDispenserBot: The creation of an automated particle delivery system that integrates with the swarm to automate the acoustic levitation process.
4. Collective Dynamics: An exploration of how individual AcoustoBots can operate independently or collaboratively to achieve shared goals, such as forming a "swarm of haptics" or a "swarm of audio."

Results
The technical evaluation focused on the accuracy and precision of the system's mechanical and acoustic performance:

• Hinge Actuation: The hinge system demonstrated a mean position error of ±2.50 cm and an orientation error of ±1.00 degree. While vertical positions showed slightly higher variance than horizontal or sloped positions, the authors note these errors are within acceptable limits given the strong focal points of the acoustic traps (optimal distance ~5 cm).
• BeadDispenserBot: The automated particle delivery achieved a mean position error of ±0.45 cm and an orientation error of ±1.00 degree. In 10 trials, the system successfully levitated particles in 7 out of 10 attempts, with the failures attributed to minor positioning variances.
• Acoustic Performance: Pressure field measurements (using microphones) were compared against MATLAB simulations.
  • Independent Haptics: Measured pressures (2950 Pa) were lower than simulations (4470 Pa) but consistent between units.
  • Collaborative Audio: Measured pressures (2665 Pa) closely matched simulations (2791 Pa), indicating effective modeling of source interactions.
  • Cooperative Levitation: Joint focus generated the highest pressures (measured ~5675 Pa vs. simulated ~7106 Pa). The authors attribute the deviation to nonlinear or environmental damping effects at higher pressures but confirm the focal point location accuracy.
  • In all scenarios, the Full Width at Half Maximum (FWHM) was approximately one wavelength, indicating tightly focused acoustic beams.

Significance and Claims
The paper claims that AcoustoBots represent a significant step forward in extending the interaction space of acoustophoretic applications. By moving away from static, singular units, the system offers a flexible, scalable, and modular platform capable of adapting to user interactions and specific scenarios in real-time. The integration of mobility, dynamic orientation, and automated particle delivery allows for novel augmentations, such as a swarm of haptics, audio, and levitation interactions that can follow users or collaborate on tasks.

The authors position this work as laying the groundwork for future deployment in larger robotic swarms. They emphasize that while the current experiments utilized two robots, the architecture is designed to support a fully distributed swarm. The significance lies in the potential to create immersive, room-scale mixed-reality experiences where users are not constrained by fixed interaction zones, enabling applications in industrial automation, museums, and collaborative workspaces. The paper concludes modestly, acknowledging that while the current setup demonstrates the feasibility of the concept, future work is required to address challenges in decentralized control, clock synchronization, and the integration of additional sensory modalities like taste and smell.