AiVibeSimultaneous Localisation and Mapping (SLAM)
SLAM is the technique by which a robot builds a map of an unknown environment while simultaneously estimating its own position within that map, using sensors such as lidar, cameras, inertial units and wheel odometry.
As the robot moves, it predicts its motion from odometry or inertial data and corrects the estimate by matching sensor observations against the map being built. Probabilistic methods handle the uncertainty: early approaches used extended Kalman filters and particle filters, and many modern systems use graph-based optimisation, in which poses and observations form a graph that is optimised as a whole. Loop closure, recognising a previously visited place, corrects accumulated drift.
SLAM underpins AMR navigation, robot vacuum cleaners, drones, autonomous vehicles and augmented reality. In factories and warehouses, an AMR is typically driven around once to create a map, which is then used for localisation during operation and updated as the layout changes.
Lidar SLAM is robust indoors, while visual SLAM uses cheaper sensors but is sensitive to lighting and texture. Long featureless corridors, glass, crowds and changing pallet layouts can cause localisation errors. Map quality and localisation confidence are monitored in operation, and SLAM-based navigation does not replace safety-rated protective sensors.
Key points
- Builds a map and localises the robot within it at the same time
- Uses lidar, cameras, inertial sensors and odometry
- Loop closure corrects accumulated drift
- Featureless or changing environments challenge localisation
Where AiVibe comes in
AiVibe designs and manufactures the AiAmbA AI Factory, whose edge devices and AI agents let people talk to robot controllers in plain language. Robotics perception is an AiAmbA use case, and robot safety functions follow ISO 10218 and never depend on the AI layer.