Training methods for an industrial robot vision system

During training period, various objects are made familiar to the vision system of the industrial robots. The extracted feature values of these known objects are stored in the vision system, and then compared with the feature values of unknown objects. Generally, there are two different methods for recognizing the objects in the industries such as:

  • Template matching methods: In the training period, a model template for the known objects is programmed in the vision system. The template includes the features of an object like diameter, area, and so on. Subsequently, these features are compared with the new image of an object. This object recognition method leads to the reduction of errors in the classification operation. Moreover, this procedure will not be handy when a lot of model templates are required to be done.
  • Structural methods: This pattern recognition method is used for considering the relationship between the edges of an object. The syntactic pattern recognition is one of the most commonly used procedures in the structural methods. In this process, an object is found as a rectangle when the image of an object is sectioned into four straight lines at right angles.

In the training session, several important conditions like part positioning, camera placement, lighting, and aperture settings should be considered. Some of the high-level programming languages like RAIL (Robot Automatic Incorporated Language), C language, etc. are used as application software in the vision system. C language was developed and used by Object Recognition System Inc., and RAIL language was developed by Automatix Inc. for programming both robot and automated vision system.

You can leave a response, or trackback from your own site.