Welcome

This a repository containing supplementary files for our project Fabrication-in-the-Loop Co-Optimization of Surfaces and Styli for Drawing Haptics published on Siggraph 2020. More information on the project, as well as, the full paper are available on our project webpage.

main.m

Example usage of the optimization with our dataset

features.mat

Dataset of measured drawing tool and surface designs for our project. The stylus nib parametrization is a pair of [radius, material] and the surface parametreization is defined by [manufacturing method, mixing ratio, mean frequency 1, deviation of frequency 1, mean frequency 2, deviation of frequency 2, layer scaling]

BayesianOptimization.m

Implementation of the optimization using Gaussian Processes. The input is a set of manufactured digital designs with observed percieved friction and vibration, a set of available stylus nibs, a description of currently printed tool (a vector with [estimated printing time in seconds, expected improvement, and nib parameters]), and the desired target haptic feedback represented as percieved friction and vibration. When executed the code suggests new tool and surface design to be manufactured

expected_improvement2D.m

Reference implementation of the analytical improvement function for two-dimensional distance minimization for Gaussain Processes. The function takes as input predicted friction and vibration with confidence bounds, target friction and vibration, distance to closest design from the target and a parameter that modulates the degree of exploration (in all of our experiments we use zeta=0.1). The output is the expected improvement towards target desired behavior.

GeneratePatternFrequency.m

A sample implementation of a freuqency-based Gabor filter that takes the input surface parameters and generates a binary mask for single-layer printing processes.

fit_GP_proxy.m

Example of Gaussian processes for predicting perceived friction and vibration using the reference implementation in Matlab.

maximize_expected_improvement_genetic.m

Genetic algorithm style optimization of new drawing tools and substrates. The function either optimizes for the stylus parameters with a set of fixed surfaces or for the surface parameters with a set of fixed stylus nibs.

Measurements

The friction and vibration measurements of each datasample captured by the device from our previous project Perception-Aware Modeling and Fabrication of Digital Drawing Tools. More details available on the project webpage.

Citation

If you use any of these materials, please, reference our original paper:

Fabrication-in-the-Loop Co-Optimization of Surfaces and Styli for Drawing Haptics
Michal Piovarči, Danny M. Kaufman, David I.W. Levin, Piotr Didyk
ACM Transactions on Graphics 39(4) (Proc. SIGGRAPH 2020, Washington DC, USA) 

bibtex:

@article{Piovarci2020,
  author = { Michal Piovar\v{c}i and Danny M. Kaufman and David I.W. Levin and Piotr Didyk},
  title = {Fabrication-in-the-Loop Co-Optimization of Surfaces and Styli for Drawing Haptics},
  journal = {ACM Transactions on Graphics (Proc. SIGGRAPH)},
  year = {2020},
  volume = {39},
  number = {4}
}