drake-new-code

Footstep Planning Code for Drake

Python code written for the new Drake distro. Before using, source setup.sh. Then python setup-test.py to make sure pydrake and mathematical programming work correctly.

Schedule

Fall 2017

• Create basic 2D footstep planner that can create a path from a goal position to end position in unconstrained space
• Make the reachable regions from footsteps into general convex regions
• Add a nominal and reachable distinction (so it is much costlier to step outside of some scaled-down version of the reachable region)
• Add an option to make the reachable region into multiple convex regions

IAP 2018

• Use LCM to feed plans into the old MATLAB simulator (runAtlasWalking.m) to allow for simulation of created footstep plans.

Spring 2018

• Tune constants to make footstep plans reasonable for Atlas
• Link in COUENNE Solver to solve angle constraints

Project Components

multiple-footsteps

multiple-footsteps/FootstepPlanner

A Python package that can create footstep plans based on various input parameters, including start position, end position, convex obstacle-free regions, and nominal footstep regions. See the README in the folder for more info.

convex-region-optimization

2D-qhull.py

Can solve for the closest point within a number of 2D regions to a goal point (run with -s to save the outputted image).

3D-qhull.py

Can solve for the closest point within a number of 3D regions to a goal point (run with -s to save the outputted image).