The BMW plant in Landshut has one of the most modern light metal foundries in the world. Whether three-, four- or six-cylinder engines: The crankshaft housings for all models in the BMW fleet are manufactured here. Previously, quality testing was carried out using a 'Rhönrad'. The crankcase was rotated using muscle power together with the 'rhönrad' in order to inspect it for defects from all sides. The problem with this is that the wheel is heavy. In addition, tall workers in particular had to bend down during each inspection process, which is detrimental to back ergonomics.

This was the decisive reason for BMW's plan to automate the quality assurance process as far as possible. As far as possible, because quality assurance still requires the trained eye of the employees. Consequently, the order placed with Augsburg-based system integrator MRK-Systeme was to develop a system in which humans and robots work together directly. The ultimate goal was an application that relieves the strain on employees, is easy to operate and meets the short cycle times. "BMW's requirements were predestined for the use of robots, as they can be used very flexibly. In the event of component changes, they can also be easily reprogrammed using point or logic correction with the Kuka SmartPAD," explains Mohre, who is responsible for Operations at HRC Systems and in this role played a key role in the installation of the automation. In short, the human-robot collaboration was intended to provide ergonomic relief for BMW employees.

For one and a half years now, two identical applications, each with a Kuka robot for payloads of up to 210 kg from the KR 210 R2900 prime K series (the K stands for the console version of the robot), have been in operation on the production line in Landshut, Bavaria. One employee at a time controls the movements of the robot using a joystick to check the components for defects in a comfortable posture.

... then check the components for faults in a comfortable posture.

© Kuka

All the crankshaft housings produced are first delivered to the work area via a conveyor belt. Depending on the type - i.e. three, four or six cylinders - the higher-level control system selects the appropriate program to control the robot's movements. The respective programs differ in the position of the gripping point, which is shifted in the longitudinal direction depending on the crankcase. This ensures that the contour jaws of the gripper can close positively. When the crankshaft housing reaches the robot's working area, the operator controls its movements using a joystick. "During the development of the joysticks, it was important to program the channels continuously and to design the direction of deflection in an intuitive arrangement for the operator," explains Mohre. The robot moves along a predefined path to the housing, closes the gripper and hands the component to the worker. To ensure that the robot does not leave the specified path or movement path, fixed spatial points were programmed into the Kuka controller with the aid of the hand-held programming device, which are defined by means of a coordinate starting from the robot origin.

Now the worker checks the component for errors using a cold light lamp. Meanwhile, the robot holds the component, which weighs around 30 kg, securely at the optimum height for the worker. Once the worker has completed the inspection process, he moves the robot and housing back to the conveyor belt and places the component on it. The interactive inspection process is completed at the touch of a button. The employee then documents whether the component is 'OK' or 'not OK'.

To ensure the safety of the employee during direct cooperation between man and machine, the system concept is based on three independent safety devices:

• a programmable logic controller (Safe PLC), which also safely detects and controls the gripper with its open/close movement, brake test and component detection functions;
• a safe robot controller that only allows the robot to run at certain speeds in defined workspaces and
• an enabling switch, which must be designed in three-stage two-channel technology in accordance with the robot product standard.

Simple operation is a must

Simple operation is also crucial for practicable human-robot collaboration (HRC). During implementation, care was taken to ensure that both right-handed and left-handed users can easily control the robot. This was achieved through the linear arrangement of three joysticks on a two-hand control. In order for the robot to move, two 'joystick feeders' must always be operated - another safety precaution. Furthermore, the control of the robot is intuitive: If the employee pushes or pulls the joystick handle forwards or backwards respectively, the robot moves in a parameterized direction. If he pushes it to the left or right, the robot also reorients itself accordingly. A thumbwheel can also be used to adjust the height of the crankcase relative to the worker. Mohre explains: "The joystick was controlled using the Kuka real-time interface RSI (editor's note: RobotSensor Interface). The signals or analog channels are interpreted by the joysticks at 250 Hz or in an interpolation cycle of 4 ms and the corresponding robot movement is then executed according to the programmed RSI project."

Expansion of the system in planning

The two systems have now been in operation since 2015. Time to draw a conclusion: "We were able to meet expectations across the board. The cycle times can be adhered to and the work is now much more pleasant for the employees," says Mohre. He and his colleagues are now even working on expanding the system. In the future, another two test benches are to be set up for quality assurance in order to keep pace with increasing production.

Safety has 'Priority 1

Dr. Albrecht Hohne, Director Human Robot Collaboration at Kuka Roboter

© Kuka

How does a standard industrial robot become HRC-capable? Dr. Albrecht Hoene, Director Human Robot Collaboration at Kuka Roboter, explains.

Mr. Hoene, what exactly makes an HRC robot?
Hoene: With HRC, safety is the first priority. To ensure this, there are two main types of collaboration: collision avoidance and collision control. In collision avoidance, it must be ensured that the robot comes to a safe standstill before it comes into contact with a human. In the case of collision control, care must be taken to ensure that biomechanical limit values - such as forces or pressures - are not exceeded in the event of a collision. The limit values were determined by a neutral body in collaboration with doctors and are naturally within a range that does not cause injuries.

Can a standard industrial robot also be 'converted' into an HRC robot?
Hoene: In principle, yes. In this case, collision avoidance is achieved by sensors that detect human access to the robot's work area - for example, safety-oriented cameras, scanners or step mats. These send a signal to the robot controller that slows or stops the robot. Sensors on the robot surface that send a corresponding signal to the robot controller when approached or touched are also conceivable. Measures for collision control include rounding off edges, avoiding externally routed cables and hoses or padding critical points.

What are the limits when it comes to making an industrial robot HRC-capable?
Hoene: When it comes to collision control, there are four variables that influence the HRC capability of an application: the moving mass of the robot and workpiece, the speed of movement, the sharpness of objects and the parts of the human body that could come into contact with the robot. It must also be taken into account that tools mounted on the robot sometimes carry out processes that prohibit HRC operation. These would be welding, milling or soldering, for example.