A*SLAM: A Dual Fisheye Stereo Edge SLAM
A*SLAM® system features combining two sets of fisheye stereo cameras and taking the image edge as the SLAM features.
The dual fisheye stereo camera sets cover the full environmental view of the SLAM system. From each fisheye stereo image pair, we can directly extract the Panorama Depth Image for initializing the SLAM feature.
The edge feature is an illumination invariant feature. We present a method of the edge-based SLAM process using both the normal and inverted images interchangeably.
To watch the origianl video, please click HERE. (Recommended to download the video file to the local disk, and watch at the 1:1 scale.)
Requisites for Robust Visual SLAM
We believe a robust visual SLAM system needs to satisfy at least the following four conditions.
- Stereo Vision
- Wide Field of View
- Full View
- Illunimation Invariant Feature
To the best of our knowledge, A*SLAM® is the only system in the world to embrace all these challenging preconditions of a robust visual SLAM system. For more details, please check this PDF file.
In many real applications, the mapping is only needed at the beginning phase. Once the map is given, the robot will only run under the localization scenarios.
Astar developed the SSML solution which enables seamless switching from the Stereo SLAM mode to the Monocular Localization mode, makes the costs of both the camera sensor and the computing device low when deploying multiple service robots in an environment and sharing the same map data.