Phoenix Mobile Experimental Platform

Research Group Robotics & Process Control

Visual Bar Code Recognition

Door Plates and directories help to facilitate human orientation in monotonous indoor environments. In order to provide similar self-localization support for autonomous mobile robots, we developed a camera-based technique for the detection and decoding of bar code labels (photo).

For computers, bar code labels are much easier to detect in camera images than natural writing. In addition, bar code labels, due to their structure, provide information redundancy concerning limited partial occlusion, rotation or perspective distortion.

There are basically three simultaneously applicable modes of using bar code labels in the service robot domain:

• Support of self-localization by marking certain coordinates in the robot's environment model (environment map) with unique bar codes.
• Providing relevant environmental spots (docking stations, lifts, etc.) with semantic information by encoding their meaning in attached bar code labels.
• Facilitation of the recognition of objects which are relevant for the robot's current task (transport boxes, etc.) by labelling with special bar codes

Implementation: Torsten Gattung

Publications: