Learning Set Representation using Kernel Functions
Generate a function whose level set gives a boundary between the given in- and out-samples using a linear optimizer.
Guide: Prof. Patricio Vela, Georgia Institute of Technology.
- Developed new theoretical guarantees for the existence of a solution for the hard-in and hard-out case.
- Implemented an algorithm in C++ that generates points and computes the representation for path planning in 11ms.
- Extended the formulation to sets in SE(2) and demonstrated utility for the representation of reachable poses for the end-effector.
Generation of in- and out-samples, followed by the linear optimization problem to obtain the level representation.
This project is still in-progress.