Efficient tracking of 3D-robot positions by dynamic triangulation

Abstract

In many industrial applications robots must accurately follow paths in 3D space. Several methods are known to measure a robot's tracking quality. For dynamic measurements, fast laser tracking systems and inertial measurement technologies have been developed. If only sequences of quasi-stationary 3D path points are to be measured, static optical triangulation is an appropriate method to check the accuracy of a robot system. Optical triangulation is based on the measurement of angles and distances in reference to a fixed baseline. Conventional triangulation methods require control systems which move scanners until a light-beam-polygon closes. It is technically complicated to servo-control moving masses and it makes the triangulation systems slow and expensive. The paper presents a new concept of dynamic triangulation, which solves 3D measurement tasks in a faster and more efficient way. The key is to replace heavy servo controlled scanning elements by a set of lighter systems which rotate freely, oscillate or sweep randomly. The robot carries a small planar sensing device to detect the spots of two scanning beams. Each time the sensor receives a pair of beam spots, a triangulation measurement is performed by evaluating the data of the stationary scanner systems and the sensor. This is none by off-line computation. The paper discusses the concept of dynamic triangulation, and makes considerations for a technical implementation.