Initially a robot have no idea where it is, how it is oriented in space and what an environment looks like. This means that a robot will have to explore its initial position, recognize useful landmarks and use them to get certainty about its localization.
After initial localization step it will be possible to command a robot to got to a point in the unexplored space, defined relative of robot "home" location.
Robot has no idea about its location, environment and position of ultra sonic sensor |
Robot got some data from radar that could be used as constraints to estimate possible location. Still, many locations are possible, so robot isn't localized yet. |
While moving to defined point a robot should detect and avoid any obstacle, build environment map and use it for localization purposes.
SLAM demo:
Some SLAM implementations
SLAM demo:
Some SLAM implementations
- Reference implementation of Sebastian Thurn's group's Fast SLAM algorithm in Java
- EducationalRobots project on Google Code (contains Fast SLAM implementation)
Further resources
- ROS (Robot Operating System) project
- OpenSLAM set of SLAM related projects, algorithms and implementations
- MyRobotLab project on Google Code
- Bashir's Robotics Vision project Using a robot equipped with video camera to generate 2D environment map
Could you please fix the link of Bashir's Robotics Vision project?
ReplyDelete