# **VISHAC** This repository includes data for *Efficient Visuo-Haptic Object Shape Completion for Robot Manipulation* paper. For code, please head to our [Github page][1]. **Authors**: Lukas Rustler, Jiri Matas, and Matej Hoffmann **Maintainer**: Lukas Rustler, lukas.rustler@fel.cvut.cz ![Schema][2] # Data description - [evaluation.zip][3] - contains Python3 Pickle (.pkl) files with computed Jaccard similarities and Chamfer distances. - Dictionary with keys: - *objects_names* - contains names of objects used in each experiments - dictonary; keys are names of the objects concatenated with "," - *e.g.,* 001_chip_can or 003_cracker_box,021_bleach_cleanser - value for each key is list with length corresponding to number of repetitions (3 in our data). Each element in the list contains file names of the used objects - *jaccard* - contains Jaccard similarities - *chamfer* - contains Chamfer distances - both both of the above applies: dictionary; keys are names of the objects concatenated by "," - value of each key is list with length corresponding to the number of repetititons (3 in our data). Each element in list contains another list, where each element's legth corresponds to he number of touches + 1 (0 touches also evaluated) and contains Chamfer/Jaccard values for reconstructions after given touch. - [logs.zip][4] - contains logs from experiments - the files is a CSV with ";" separated values. The columns are: - timestamp - timestamp of start of the experiment - folder for meshes etc. are named with this timestamp - objects - comma separated name of the objects used - (number of reconstructions, max_time,number of repetitions) - repetition - given repetitions on the given object - 1 to number of repetitions - time - time needed to completion - [exps.zip][5] - settings for easy experiment running - json files with mandatory fields: - "objects" - list of list; each inner list includes names of objects used, separated by comma - "origins" - list of list; each inner list include 3 elements list with x,y,z coordinates of the center of the objects; mainly for simulation - "repetitions" - list if ints; each integer stands for number of repetitions to make on the given objects - "reconstructions" - list of ints; number of touches to done for each object; - "max_time" - list of floats; maximal time for the pipeline to run - all the fields must have the same length - [pcd.zip][6] - contains point cloud related files - each folder is named with timestamp correspoding to the start of the experiment - each folder contains multiple folders named "repX", where X is the number of touches - each subfolder contains ".pcd" files for each object in the given experiment, normals for the point cloud, colors, touches and normals for touches. - [npy.zip][7] - contains numpy files for the objects - each folder is named with timestamp correspoding to the start of the experiment - each subfolder contains points from point cloud in ".npy" file, the center of the point cloud and the scale of the point cloud - in the root of each folder are also two files with times for touch/reconstruction - [meshes.zip][8] - contains mesh related things - each folder is named with timestamp correspoding to the start of the experiment - each subfolder contains multiple versions of the mesh in world or RVIZ coordinate frame. - Also, "_eikonal.npy" files with the output shape of the netowrk with uncertainty is available - the file is Nx4 numpy array and the first columns is uncertainty and the rest arer 3D coordinates of individual points - [rosbags][9] - rosbags collected during the experiments - rosbags for real experiments contain joint states of the robot - rosbahs for the simulated experiments contain joint states of the robot and depth and rgb images (1 every 2 seconds) [1]: https://github.com/ctu-vras/vishac [2]: https://raw.githubusercontent.com/ctu-vras/vishac/master/readme_figs/schema.png [3]: https://osf.io/j6rkd/files/osfstorage/6408a98086049004ed08693e [4]: https://osf.io/j6rkd/files/osfstorage/64085a80e25636048e231f17 [5]: https://osf.io/j6rkd/files/osfstorage/64085a81e256360491231848 [6]: https://osf.io/j6rkd/files/osfstorage/64085b29c74723049d10b1e4 [7]: https://osf.io/j6rkd/files/osfstorage/64085b9d86049004a1086f3e [8]: https://osf.io/j6rkd/files/osfstorage/64087dae86049004c308738c [9]: https://osf.io/j6rkd/files/osfstorage/640869b0e256360497231d26