ºÚÁÏÍø

The title of the thesis: Manipulation and assembly of objects using spatially nonlinear stochastic force fields 

Thesis defender: Artur Kopitca
Opponent: Prof. Antoine Ferreira, INSA Centre Val de Loire, France
Custos: Prof. Quan Zhou, Aalto University School of Electrical Engineering

Spatially nonlinear stochastic force fields, ubiquitous in nature, hold remarkable yet underexplored potential for shaping, transporting, and assembling objects. Winds, for instance, with random and nonlinear variations in speed and direction, form complex sand dunes and carry diverse objects across long distances. Inspired by such phenomena, this thesis explores the controlled use of two representative fields—vibration and airflow—for object assembly and remote manipulation. Though differing in physical origin, both exhibit nonlinear spatial variations in force strength and direction, coupled with stochastic fluctuations. 

The first part demonstrates that acoustic vibration fields can be programmed to assemble particles into target 2D shapes on a vibrating plate, surpassing prior work constrained by intrinsic nodal patterns. A data-driven model predicts stochastic particle motion, and an optimization algorithm iteratively applies the field to achieve desired shapes. This enables the assembly of up to 100 particles into recognizable forms such as letters and geometric figures—echoing natural processes driven by long-term, nonlinear, time-varying stimuli and introducing a new paradigm for field-based assembly. 

The second part explores airflow fields for long-range, non-contact manipulation of diverse objects. Unlike conventional methods (e.g., magnetic, acoustic, or optical) limited by material type or range, a single jet-induced airflow field can automatically guide objects up to 2.7 meters away with high precision (≤1.5 cm error), on both solid and water surfaces, in the presence of obstacles or disturbances, and across various application cases. Control is achieved via two strategies: a feedback-based model-free controller and a model-based controller using an analytical airflow model and learned object dynamics for multi-object manipulation. 

This work bridges nature-inspired and engineered approaches to manipulation, highlighting the potential of stochastic force fields for applications in robotics, manufacturing, and beyond.

Key words: Chladni plate, airflow field, motion control, machine learning

Thesis available for public display 10 days prior to the defence at .