LiDAR

LiDAR is what lets AMRs know where they are and what's around them without relying on floor infrastructure. The sensor sweeps the warehouse or factory floor continuously, producing a live picture of the robot's immediate environment accurate to within a few centimetres.

That precision matters in a busy warehouse. A robot moving near a forklift or a pallet that's been left in an aisle needs to detect the obstacle and respond quickly. LiDAR gives the reaction time to do that safely, and it does it without requiring any changes to the facility.

Most AMRs use 2D LiDAR, which maps a horizontal plane at robot height. The limitation is low objects, a cage left on the floor or a pallet at a height the sensor sweeps past will be found by the sensor but anything above may be missed. Higher-spec systems add 3D LiDAR or depth cameras to close that gap. It's worth asking vendors specifically about low-obstacle detection in their demo, not just general navigation.

The LiDAR sensor unit sits on the AMR and rotates continuously, firing laser pulses across a wide arc. Each pulse travels until it hits something, then the reflected signal returns to the unit. The time between sending the pulse and the return gives the distance to the object. Thousands of pulses per second produce a dense point cloud, giving a real-time map of everything around the robot.

The SLAM algorithm uses this point cloud to match the current sensor picture against the stored facility map, calculating the robot's exact position. When something new appears, it shows up immediately in the point cloud and the navigation system routes around it.

Sensor specifications that matter in practice are field of view (angular coverage), detection range (how far the sensor reliably reads), and update frequency (how quickly the map refreshes). These affect performance most in large facilities or environments with complex or unpredictable obstacles.