For two centuries the assumption was that a Chladni plate does one thing well — collect particles on nodal lines — and that everything happening between those lines is noise. Quan Zhou's group at Aalto University tested that assumption and found the out-of-nodal-lines motion regular enough to model statistically, then to predict, then to control.
With a single acoustic actuator and a carefully chosen sequence of tones, they moved multiple objects independently: separate trajectories, pattern transformations and sorting, in electronic components, water droplets carried on solid substrates, plant seeds, candy balls and metal parts.
The practical lesson for anyone building plate rigs is that the interesting behaviour is not only at resonance. The sequence of notes matters as much as any single note, and the plate is closer to a programmable surface than a static pattern generator. Open access in Nature Communications.