Is That a Chair? Imagining Affordances Using Simulations of an Articulated Human Body (ICRA2020)
Hongtao Wu, Deven Misra, Gregory Chirikjian
- Pybullet (3.0.8): Simulation platform
pip install pybullet==3.0.8
- Trimesh (3.9.1): Used for measure geometry of the object
pip install trimesh==3.9.1
- Miscellaneous
pip install numpy==1.16.6
- V-HACD: Build the library from the V-HACD repo
git clone https://github.com/kmammou/v-hacd cd v-hacd/src/ mkdir build && cd build cmake .. make
- meshlab (2020.03): Used for getting the physical information of the scanned object. Latest version of meshlab will not work with xml file. It needs python 3.6 (Pymeshlab) to interface with. Download the source code from the Release page. Clone the vcglib after unzip the code. Remember to checkout to the correct commit meshlab 2020.03 was using.
Then go to the root directory and use the bash file in /install to install.
cd <PATH_TO_MESHLAB> git clone https://github.com/cnr-isti-vclab/vcglib.git cd vcglib git checkout 5fa560e9e6
The executable of meshlab and meshlabserver will be installed in /distrib. We will be using the meshlabserver in the code to compute geometric metrics.cd ../ bash install/linux/linux_setup_env_ubuntu.sh bash install/linux/linux_make_it.sh
-
Specify the data folder path, result folder, v-hacd path and meshlab path in
./src/chair_imagination_clasification.py
-
Run the experiment.
python src/chair_imagination_clasification.py
If you find this code and/or the data useful in your work, please consider citing
@inproceedings{wu2020chair,
title={Is that a chair? imagining affordances using simulations of an articulated human body},
author={Wu, Hongtao and Misra, Deven and Chirikjian, Gregory S},
booktitle={2020 IEEE International Conference on Robotics and Automation (ICRA)},
pages={7240--7246},
year={2020},
organization={IEEE}
}