Intelligent Approximation of the C-Space in Motion Planning

GRAND Seminar Friday, October 12th, 11 am, Room: 4201

Chinwe Ekenna
Assistant Professor
Computer Science
University at Albany, SUNY

Abstract:

With the continuous improvement of the capabilities of robots and the increasing complexity of the environments they successfully traverse, my research is geared towards the investigation of mathematical concepts and algorithms about the heterogeneous nature of these environments within the context of motion planning. My research will look into existing properties about planning spaces such as visibility, expansiveness and homotopy class and how they can be generalized within the context of heterogeneous spaces. We will also introduce new properties that specifically better define heterogeneous properties. We have recently developed an algorithm called the Rapidly Exploring Random Search Explorer (RESE), which extends ideas from Rapidly-exploring Random Graphs (RRGs) and includes a Q-learning machine learning technique to help track witness nodes of successful trajectories while also attempting connections in narrow regions by considering multiple directions. These algorithms are built within the framework of a sampling-based motion planning which includes a group of algorithms that sample random points in the environment. My research has the potential to change the way planning spaces are planned and help intelligently define how algorithms are built and utilized given any environment.

Short Bio:

Dr. Ekenna’s research centers on intelligent motion planning applied to robotics and proteins. She has explored intelligent adaptation of robotic motion planning to improve planning time and the utilization of available methods to improve overall quality of results. Her research interest includes Robotics, Machine learning, and Computational Biology.