Work Area Monitoring in Dynamic Environments Using Multiple Auto-aligning 3D Sensors

Compared to current industry standards future production systems will be more flexible and robust and will adapt to unforeseen states and events. Industrial robots will interact with each other as well as with human coworkers. To be able to act in such a dynamic environment, each acting entity ideally needs complete knowledge of its surroundings, concerning working materials as well as other working entities. Therefore new monitoring methods providing complete coverage for complex and changing working areas are needed. While single 3D sensors already provide detailed information within their field of view, complete coverage of a complete work area can only be achieved by relying on a multitude of these sensors. However, to provide useful information all data of each sensor must be aligned to each other and fused into an overall world picture. To be able to align the data correctly, the position and orientation of each sensor must be known with sufficient exactness. In a quickly changing dynamic environment, the positions of sensors are not fixed, but must be adjusted to maintain optimal coverage. Therefore, the sensors need to autonomously align themselves in real-time. This can be achieved by adding defined markers with given geometrical patterns to the environment which can be used for calibration and localization of each sensor. As soon as two sensors detect the same marker, their relative position to each other can be calculated. Additional anchor markers at fixed positions serve as global reference points for the base coordinate system. In this paper we present a prototype for a self-aligning monitoring system based on ROS and Microsoft Kinects. This system is capable of autonomously real-time calibrating itself relatively and in respect to a global coordinate system as well as to detect and track defined objects within the working area.