Acoustic levitation provides a promising method for contactless transportation of objects in air. This opens opportunities, particularly in high-precision industries where mechanical contact is undesirable. For acoustic levitation to serve as a viable alternative, it is necessary to investigate the extent to which the operating speed can be increased while maintaining precise control.
Therefore, this thesis investigates the properties on which the acceleration and velocity limits depend and experimentally evaluates their influence. The study consists of two parts: an acceleration analysis and a velocity analysis.

In the first part, the maximum stable acceleration is set by the acoustic trap stiffness and expressed in terms of the measured eigenfrequency \(f_0\) and the wavelength \(\lambda\). Experiments are conducted on spherical particles of varying diameters (0.593 to 2.886 mm) and densities (30 to 2500 kg/m\(^3\)), driven sinusoidally at sub-resonant frequencies. The measured acceleration limits are compared with theoretical predictions based on a linearised small-particle (Rayleigh-limit) force description and a Taylor-expansion-based model. For most particles, the Taylor-based formulation reproduces the experimental acceleration limits within \(\pm 15\%\) in both directions. In contrast, the linearised model systematically overpredicts the achievable acceleration, with typical deviations of 40-50\% in the vertical direction and similar overestimation in the horizontal direction. The remaining discrepancies are attributed to cross-axis coupling and gravity-induced equilibrium offsets.

In the second part, the maximum achievable velocity is studied through a balance in forces between the acoustic radiation force and aerodynamic drag, using a nonlinear drag model. Velocity tests are performed using sinusoidal excitation with gradually increasing amplitude. The measured velocities are compared with analytically derived predictions and dynamic simulations. While agreement is good in the horizontal direction, the vertical direction shows larger phase lag and earlier loss of stability, indicating increased damping and nonlinear effects at high speeds.

The results demonstrate that maximum acceleration scales with \(f_0^2 \lambda\) and is direction dependent. Another result is that the particle density and diameter influence performance via eigenfrequency and drag. For acceleration, the Gor'kov-based model shows reasonable agreement, whereas for velocity, a Reynolds number-dependent increase in damping is observed, particularly in the vertical direction.

This work provides a validated experimental methodology for characterizing acceleration and velocity limits in acoustic transport, contributing to the design of faster and more reliable non-contact handling systems.

The industry has continuously downscaled integrated circuits onto a silicon base plate, now mass producing microchips at sub-millimetre scale. As adhesive forces grow at this scale, methods of non contact handling are becoming increasingly more interesting for transporting the chips. Currently, the most promising form of non-contact handling is acoustic levitation. In this approach, cuboid shapes can be manipulated in six degrees of freedom using multi-axis acoustic levitators. Levitated objects exhibit low stiffness and low damping, making the system vulnerable to disturbances. To enable precise positioning of a cuboid chip, feedback is required to counteract these disturbances. Feedback could potentially be implemented using a low-latency tracking algorithm.
This work focusses on implementing an event camera to track the position of an acoustically levitating cuboid within a multi-axis levitator. The Metavision 3D Detection and Tracking algorithm is used to perform the tracking. First, the illumination and surface properties of various concepts are investigated to generate events along the edges of the cuboid on the event plane, while minimizing events on the cuboid’s surface. These aspects have been identified as essential for the effectiveness of the algorithm. Second, the conditions under which the tracking algorithm can be applied are examined. These include both static and dynamic conditions, failure velocity, and influence of various parameters.
The findings from these investigations demonstrate that the Metavision tracking algorithm is capable of reliably tracking a levitating cuboid in stationary conditions with multiple high-intensity diffuse light sources using an event camera, ensuring that the tracking parameters are specified within certain bounds. However, when the cuboid is actively actuated, the algorithm becomes unreliable. Frequent tracking failures are observed under these dynamic conditions. These results contribute to the broader concept of implementing feedback for acoustically levitating cuboids in potential future applications.

This thesis investigates the forces and torques acting on acoustically levitated cuboids, with the aim of improving translational and rotational control in acoustic transportation applications. Traditional trans port methods rely on mechanical pick-and-place systems, which introduce challenges such as fragility and contamination. Acoustic levitation presents a contactless alternative that eliminates these issues. To address the research question: “1: Can we develop a model to determine the stiffness and torsional stiffness values of acoustically levitating cuboids of arbitrary shape, in an arbitrary pressure field?” the study begins by reviewing the fundamental principles of acoustic levitation, including acoustic radiation force, the Gorkov potential, and phased array transducer (PAT) configurations. It then explores various modeling approaches for determining the forces and torques on levitated objects, comparing the strengths and limitations of Gor’kov-based, finite difference time domain (FDTD), finite and boundary element method (FEM/BEM), and a proposed simplified trapping model.
The findings show that while the simplified trapping model lacks predictive accuracy, a Finite Element Method (FEM) model was introduced as a more reliable alternative. The FEM model, using a sound hard boundary assumption, agrees well with prior models of forces on non-spherical particles, that cannot be predicted with Gorkov’s or King’s method, and its implementation in Comsol makes it more accessible to a broad audience. In addition, the model predicts accurately the torques that act on non spherical particles and for the first time such a model is experimentally verified for all three translational and three rotational degrees of freedom....

systems. Miniaturization creates challenges in handling delicate, small components, rendering traditional methods inadequate, necessitating an alternative. Acoustic levitation, which uses ultrasound waves to suspend objects without physical contact, offers a promising solution for containerless transportation. This technology can accommodate various materials regardless of their electromagnetic or optical properties and eliminates particle generation due to friction. Applications span across chemistry, biology, and high-tech industries, enabling manipulation and assembly of small components. However, acoustic levitation is highly sensitive to environmental factors such as temperature and air currents, which can compromise stability. Most studies have primarily focused on understanding the dynamics of acoustic levitation and developing models to describe the underlying physical phenomena. This study shifts towards actively controlling them by adding artificial damping. To achieve this, this thesis introduces the use of event cameras, which capture high-speed movements without redundant frame data, making them highly efficient. The goal is to create a control feedback loop to correct for disturbances and improve the step response of acoustically levitated objects. The methodology involved integrating an event camera into the acoustic levitation setup and implementing a tracking algorithm to track the position of a levitated object. System identification was performed, revealing higher-order harmonics and allowing the development of a second-order model of the levitated object. This model was used to design a feedback controller integrated with the event camera. Experimental results demonstrated a substantial improvement in system performance. The bandwidth of the system improved to 40.8 Hz, the settling time decreased from 10 seconds to 0.8 seconds, an improvement by a factor of 12.5. Additionally, the steady-state error was significantly reduced, and the introduction of artificial damping successfully eliminated the first resonance peak, though a second higher-order resonance peak remains. This work highlights the potential of using event cameras in control loops for acoustic levitation, representing a significant step towards more precise control of levitated objects.

Acoustic levitation is an novel method that could lend itself very useful to fast and precise transportation of small objects. To verify the acoustic field and the modelling assumptions, the acoustic field must be visualised and quantified. Different methods to visualise a pressure gradient are investigated and rated against design requirements. These requirements are: 1. difficulty, as the setup should be buildable within the period of a MSc. thesis. 2. Time per measurement and 3. measurement sensitivity. It is decided that using a double coincidence spherical mirror schlieren setup is most suitable measurement method. To capture the high speed sphenomena of ultrasound acoustic waves in air, the schlieren light source is pulsed and matched to the frequency of the acoustics to ’freeze’ the acoustic wave in the air. Parameters that influence the schlieren image are identified and studied. These parameters are duty cycle, cut-off, ISO, light source phase delay and light source frequency. A horizontal cut-off is determined as the best, while it is determined that the other parameters can be adjusted dependent on the acoustic field. The acoustic field generated by double phased array is observed. Trap locations and types can be identified using the schlieren system. Quantification of the schlieren images has been attempted but without satisfactory results, mainly due to an interesting observation regarding a correlation between the camera focus distance and the pressure gradient. The schlieren system is a useful research tool and successful at visualising
the pressure gradient under the assumption that it is constant over the optical axis. Further research is needed to for quantification and to assess the impact of the observation.