ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
1 | initial version |
This depends on your sensor setup. If it can provide 3D-pointclouds at a high enough rate, you can do 3D-SLAM by aligning the pointclouds with an ICP-Algorithm. You can also use the pioneer's wheel odometry to replace the IMU to some extend.
2 | No.2 Revision |
This depends on your sensor setup. If it can provide 3D-pointclouds at a high enough rate, you can do 3D-SLAM by aligning the pointclouds with an ICP-Algorithm. You can also use the pioneer's wheel odometry to replace the IMU to some extend.
Edit: A good ICP-Variant is the GeneralizedIterativeClosestPoint from pcl-registration library.
To create a 3D-Map with a 2D-Sensor (like your Hokuyo), you have to mount the Scanner on a tilt-unit to periodically tilt it up and down. Then PCL can create a pointcloud out of all scans from one full sweep of the scanner.