Learning Planning Spaces

PyTorch implementation of LearnedSamplingDistributions by Brian Ichter, James Harrison, Marco Pavone

Here, the focus is on learning focused search spaces for path planning applications

Ichter et al. Github link

Their paper: https://ieeexplore.ieee.org/abstract/document/8460730

Ichter, B., Harrison, J., & Pavone, M. (2018, May). 
Learning sampling distributions for robot motion planning. 
In 2018 IEEE International Conference on Robotics and Automation (ICRA) (pp. 7087-7094). IEEE.

Examples

Linear Narrow Passages: reproducing the example by Brian Ichter, James Harrison, Marco Pavone. Given an occupancy grid with small passages and a start and stop location, predict a focused search space for shortest-distance planning

• Notebook: LinearNarrowPassages_Train.ipynb
• Ichter et al.'s original (tensorflow): https://github.com/StanfordASL/LearnedSamplingDistributions

Boston Harbor: given start, goal locations in a static environment, predict a focused search space for shortest-distance plannng

• Notebook: BostonHarbor.ipynb
• Note: data too large for GitHub. Will upload somewhere soon and provide link

Boston Harbor Water Currents: given start, goal locations and water current predictions, predict a focused search space for energy-efficient planning

• Coming soon!