Stereo Vision SLAM for a Robotic Wheelchair

Introduction

The intended application of this research is to integrate stereo vision and SLAM to create a map while a robotic wheelchair is navigating in its environment. The vision based mapping is one component of the assisted navigation system being developed for the wheelchair.

Approach

Simultaneous localization and mapping (SLAM) has been widely used for navigation and typically makes use of laser range-finders or sonar. We are going to use pmap, a SLAM implementation that assumes the use of a SICK laser. It will be modified to accommodate the stereo vision data.

An advantage of using stereo vision over laser range-finders is the ability to detect obstacles at different heights. Laser range-finders return distance measurements on a 2D plane at a fix height. With stereo vision we can mark the closest obstacle regardless of height. For example, when a laser range finder sees four table legs, it only marks the table legs as occupied space in the map. This could cause a robot to try to drive under a table it should be avoiding. Our stereo vision approach would be able to detect the entire table and avoid collision.

Stereo Vision Screenshots

The following are screenshots from our application that converts the disparity information into a local occupancy grid to be passed to pmap.

left input image	right input image



rectified left image	rectified right image


disparity image	confidence map

2D Maps

Maps generated using stereovision and PMAP.

sv3d Animation

We created an OpenGL application named sv3d to visualize the 3D point cloud that corresponds to the disparity map. It is an interactive real-time display of the 3D point cloud returned from SVS. Click on the animation for a high-resolution (9MB) version.