Python Implementation of Fastron

A Cython interface to implement Fastron library with Rational Quadratic kernel in Python environment. This implementation is based on the paper Learning-Based Proxy Collision Detection for Robot Motion Planning Applications . If you use this work in your research, please cite

@article{das2020learning,
  title={Learning-based Proxy Collision Detection for Robot Motion Planning Applications},
  author={Das, Nikhil and Yip, Michael},
  journal={IEEE Transactions on Robotics},
  year={2020},
  publisher={IEEE}
}

Use Fastron in Python

The Cython wrapper is already setup, hence Fastron can be directly imported. Initialze Fastron class with training data (numpy 2d array)

import numpy as np
import fastronWrapper

...

fastron = fastronWrapper.PyFastron(data)

Compile the Code

Before compiling the Fastron Cython code, make sure you have Eigen and Eigency installed.

pip install eigen
pip install eigency

If using Windows, you may need to install/update MS Visual C++ 14.0.

NOTE: before compile, please read the issues section below

To compile, run command:

python setup.py build_ext --inplace

Issues

When trying to setup the Cython wrapper, compilation error may occur on some Eigen functions. This is because the bridge between numpy and Eigen, Eigency, maintains its own version of Eigen library and it is outdated. To fix this issue, replace that library with the latest Eigen library.

cd /tmp
curl -o eigen-3.3.7.tar.bz https://bitbucket.org/eigen/eigen/get/3.3.7.tar.bz2
tar -xvf eigen-3.3.7

# Need to rename it to eigen_3.2.8;
mv <untarred_directory> <conda_env>/lib/python3.7/site-packages/eigency/eigen_3.2.8

On Windows, replace Eigency's version of Eigen 3.2.8 (in site-packages) with Eigen 3.3.7 by selecting the contents in the Eigen 3.3.7 folder and pasting them into Eigency's Eigen folder.

Usage

Import the fastronWrapper.pyx file to allow instantiation of PyFastron objects.

import pyximport
pyximport.install()

import numpy as np
from fastronWrapper.fastronWrapper import PyFastron

The PyFastron constructor only has a dataset of configurations (as a Numpy matrix) as an argument. Data labels (as a Numpy array) and kernel and model parameters can be set after construction.

The updateModel() method learns the weights for the model, and eval() predicts the collision status values with the trained model.

Example

# Initialize PyFastron
fastron = PyFastron(data) # where data.shape = (N, d)

fastron.y = y # where y.shape = (N,)

fastron.g = 10
fastron.maxUpdates = 10000
fastron.maxSupportPoints = 1500
fastron.beta = 100

# Train model
fastron.updateModel()

# Predict values for a test set
pred = fastron.eval(data_test) # where data_test.shape = (N_test, d) 

Requirements

• Python
• Cython
• Numpy
• Eigen
• Eigency

Credits

• Mingwei Xu - InspireX96
• Nikhil Das - nkhldas