================================= TrainRobots Dataset V2.0 ================================= Contact Information: =========================== Author: Muhannad Alomari Email1: scmara@leeds.ac.uk Email2: omari.1988@gmail.com Author: Dr. Kais Dukes Email1: sckd@leeds.ac.uk Description: =========================== Train Robots is a dataset of synthetic scenes of a robot arm picking up and putting down different objects together with natural language descriptions of these actions. These descriptions were obtained using Amazon Mechanical Turk. An extension to the original dataset includes two more shapes and an additional colour. Finally, the natural given descriptions are given in Arabic as well as English. Contents: =========================== 1- treebank.zip : contains the original dataset 2- scenes.zip : contains the simulated video for each of the 1000 scenes as images 3- raw_data.zip : contains the raw data for each video in the following format A- the sentenses, each sentence takes 2 lines, the first says sentence followed by it's ID, and the second says if its good or spam and has the actual sentence sentence:1 GOOD:Move the green pyramid in the botton left corner B- the objects, each object takes 6 lines, the first says object followed by it's ID, then it's x,y,z location in each frame, then its RGB colour values (note it's RGB and not HSV), and finally it's shape value on a scale from 0-1, (0=cube, .33=sphere, .66=cylinder, 1.0=prism). object:0 x:3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00, y:3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00,3.00, z:0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00,0.00, F_RGB:1,1,0 F_SHAPE:0.66 Published papers: =========================== @inproceedings{alomari2016, title={Unsupervised {G}rounding of {T}extual {D}escriptions of {O}bject {F}eatures and {A}ctions in {V}ideo}, author={Alomari, Muhannad and Chinellato, Eris and Gatsoulis, Yiannis and Hogg, David C. and Cohn, Anthony G.}, booktitle={15th International Conference on Principles of Knowledge Representation and Reasoning}, year={2016} } @inproceedings{dukes2013train, title={Train {R}obots: A {D}ataset for {N}atural {L}anguage {H}uman-{R}obot {S}patial {I}nteraction through {V}erbal {C}ommands}, author={Dukes, Kais}, booktitle={International Conference on Social Robotics (ICSR). Embodied Communication of Goals and Intentions Workshop}, year={2013} }