Skip to content

brenhertel/ElasticMaps

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ElasticMaps

Implementations of Elastic Maps for Trajectory Learning from Demonstration.

This repository implements Elastic Maps, specifically polyline Elastic Maps for robot trajectories. Two versions are implemented, using a matrix-inversion least squares approach as well as a convex optimization approach. For further details and references please see the following papers:

"Robot Learning from Demonstration Using Elastic Maps" by B. Hertel, M. Pelland, and S. R. Ahmadzadeh, available here.

"Confidence-Based Skill Reproduction Through Perturbation Analysis" by B. Hertel and S. R. Ahmadzadeh, available here.

These two methods solve the same problem in different ways. The primary difference between them is that the scaling factors (stretching and bending) are differently incorporated into the optimization. This means that given the same problem, the same parameters will arrive at different solutions for the two methods. Some trial and error may be necessary to find correct parameters (higher stretching promotes a shorter-length solution, higher bending promotes a straighter solution).

This repository implements the method described in the papers above using Python. Necessary libraries include numpy, matplotlib, and cvxpy>=1.5. Scripts which perform individual experiments are included, as well as other necessary utilities. If you have any questions, please contact Brendan Hertel ([email protected]).

About

Implementations of Elastic Maps for Trajectory Learning from Demonstration.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages