The task of this robotic control station, built on the basis of multi-camera vision system, is to control the conformity of production and assembly of the copper pipes system of the heat pump. The inspection includes:
Copper pipes location inspection - for the heat pump to work correctly, it is vital to ensure proper distance between copper pipes and other components. Using a 3D-camera the vision system is able to control the location of inspected elements in space.
Soldered connections gas tightness inspection - since the pump installation is filled with technical gas it is very important to maintain its gas tightness. Vision system is able to locate points, to which robot moves a tool which takes a sample of air from the copper elements connection area for further analysis.
Construction of heat pump control station
The station is placed on the production line of the household appliances. Its main components are listed below:
ABB Industrial robot
Vision system running on a PC computer
Siemens PLC controller
The vision system was developed in Adaptive Vision software environment and it works as an independent application on a PC computer. Two cameras, located on a robot wrist, are an essential part of this system. The communication with these cameras is performed using GigEVision and GeniCam protocols, but the TCP/IP protocol is also used for communication with the 3D-scanner. To communicate with the PLC controller another TCP/IP connection was used.
After performing the acquisition and calculating the outcome results, the results of the inspection are visualized on the monitor display and sent further to the PLC. The application also stores the history of the inspections and every inspection can be accessed anytime. It also contain an expanded module for parameters configuration, which allows for easy modification of the appropriate ranges and acceptable positions of copper pipes in the heat pump space.
Moreover, the application has an auto-diagnostic function, which continuously supervises in particular the correctness of the communication process with cameras and PLC controller. Every disturbance is signalled in the application itself as well as triggers a notification sent to the PLC controller.
The complete work cycle for this station consists of the following steps:
After the pump is stopped in the control station, the system acquires a set of data which defines the type and model of the heat pump. According to this data the vision system chooses the appropriate set of parameters for performing further quality inspection.
During the first robot ride along the inspection area, the 3D-scan is acquired. Next, the vision system specifies on its basis the location of the heat pump and calculates its full orientation in the 3D space. This way, the algorithm is based not on the rigid patters of calculations in global coordinate system, but on the local coordinate system of the inspected element. This solution removes the necessity to have the perfect positioning of the inspected element. Translations of a few centimetres and rotations do not cause any problems for the vision system.
From this moment onward the vision system overrides the control system of the heat pump control station and takes full control over it. The vision system establishes the position for the robot, based on the performed calculations. In this position the image is acquired by the Basler 2D matrix-camera. Based on this image the system determines if the specified space is not affected by a overly tilted capillary, which is an element of the inspected pipe system. The specificity of the capillary location made it impossible to use 3D analysis method to perform this part of the inspection. Due to the fact, that the inspection is performed on the open area, the algorithm is resistant to changes in the lightning conditions and it adjusts parameters to be failure-free.
In the next step, the robot moves back to the start position to perform the second scan. This position is calculated from the data from the first scan.
After performing the second 3D scan, the vision system locates points where the tightness quality control will be executed. Next, calculated points coordinates are sent to the robot, which performs an adequate test.
At the end of the inspection procedure, the vision system sends the analysis of the results to the PLC controller and the robot comes back to the starting point and waits for the next algorithm cycle.