Research Project P3:
Miniature SLAM system for 3-D Simultaneous Localization and Mapping

Research Project Goal

Build and develop a miniature system for 3-D environment modeling, localization and navigation. In recent years, 3-D modeling of environment has become vastly popular in the robotics community. However, today's best methods all use heavy and expensive sensors, which are prohibitive for many applications. Recently, we have received a new range finder by Hokuyo which is significantly smaller and energy efficient. The goal of this project is to develop a full 3-D modeling method that uses this sensors to build 3-D environment models, and then cross-references such models for robot navigation.

Research Project Scope

The idea is to use a miniature range finder, build a server for 1-D actuation, and mount it on a moving platform to acquire 3-D models. Then build software for acquiring 3-D point clouds, and for matching them using a variant of the ICP (iterative closest point) algorithm. Use this software to perform simultaneous localization and mapping, or SLAM.

Data

Image of the sensor (image (c) Hokuyo):



You can see the sensor in action in these two movies:

                           

Tasks

• Build actuation platform for the sensor using servo motor and shaft encoder (requires hardware skills!!!)
• Program computer interface for data acquisition; develop 3-D visualization software
• Implement scan matcher (ICP) for SLAM
• Experiment in the physical world.

Research Project Status

student names here

Point of Contact

Sebastian Thrun

Midterm Report

not yet submitted

Final Report

not yet submitted