Institut für
Robotik und Prozessinformatik

Deutsch   English

Zur Zeit leider nur in Englisch vorhanden...

Active environment concept for efficient force control systems


The considerable progress made in manufacturing has made it possible to construct industrial manipulators with more and more stiff joints. This presents an obvious advantage in an unlimited space, which results in increased positioning accuracy. However, problems arise when the robot is equipped with a torque and force sensor for the purpose of a controlled action on the environment, for example in the tasks of autonomous assembly. In such cases the high environment stiffness may account for generation of big forces, which have a detrimental effect on the mechanical construction of the manipulator and constrain the allowable robot velocities at the instants when the robot comes into contact with the environment. Therefore, passive or active compliance systems are introduced between the stiff manipulator and also the mostly stiff environment. They are usually added as a part at the end of the robot arm. A passive compliance uses some mechanical devices composed of springs, sliding axles and knee joints. In a passive compliance devices, there is no interact between the control computer of the robot and the compliant device. The chief drawback of such a solution is lack of information about the exact position of the object in the space. Since precise positioning is a matter of prime importance in the majority of applications, therefore active compliance systems equipped with sensors coupled with the robot control system take on great significance. One such commercial example is the DLR lightweight manipulator arm with torque sensors in each joint along with collision detection and safe reaction system. Although the active compliance systems based on the hybrid position-force control are much more attractive than the passive solutions based on the impedance control, nonetheless they present a great challenge due to occurring stability problems. Additionally, it should be remembered that the presence of a passive compliance between the manipulator and the environment is required for entering into contact to be non-invasive, even if an active compliance is used.

Six-joint industrial manipulators are a complex multi-input and multi-output systems characterized by strong nonlinear and time-varying properties. Additionally, if a manipulator comes into environmental contact, the problem of time-variant environmental stiffness arises. In such a case the stiffness may fluctuate even 500-fold, which results in changing the gain of the force control loop. This fact may has a dramatic effect on the settling time, if a classic low-robust singleloop PID system is applied.

The experiments have been performed using the manipulator Stäubli RX60 equipped with a JR3- manufactured sensor of torques and forces. To illustrate advantages offered by force control with an active environment, a series of tests has been initially done for the classic case, i.e. without any stiffness actuator. In using the hybrid position/force control, the tests have been carried out for two contact surfaces of different stiffness, namely an aluminum plate of high stiffness (Fig. 1), and a PVC plate of low stiffness (Fig. 2).

Fig. 1 Environment contact of high stiffness (aluminum plate).

Fig. 2 Environment contact of low stiffness (PVC plate).

Fig. 3 Force control in the case of contact environment of high stiffness.

Fig. 4 Force control in the case of contact environment of low stiffness.

The force control response obtained for these cases is given in Fig. 3 and Fig. 4. The result is not surprising. A change in the environmental stiffness amounts to a change in the gain value of the force control loop. Since the gain has been lowered many-fold, as result the settling time has undergone a significant increase. Namely, the settling time has been changed around 30-fold for the case of the low stiffness contact, if compared with the high stiffness contact.

The problems described above gave rise to an idea to apply the concept of active environmental stiffness to the synthesis of force control systems. The solution is based on a high-dynamical six- DOF controlled Hexa platform (fig. 5), which represents a contact environment for the industrial manipulator (fig. 6). The mechatronic design employed here makes the stiffness sensed by the manipulator tip invariant and freely definable for all degrees of freedom, no matter whether the manipulator comes into contact with steel or soft rubber. Therewith the full robustness to stiffness variations is achieved, which results in invariant settling times; hence, it facilitates the process of assembling.

Fig. 5 Real system.

Fig. 6 Experimental setup.

Fig. 7 Active environment and contacting a high stiffness.

Fig. 8 Active environment and contacting a low stiffness .

The stiffness of the environment to be controlled has been determined in the same way as that for the case of contacting the PVC plate by the manipulator. Two tests have been carried out; both ones with controller settings remaining unchanged: the manipulator has come into contact with active environment (Hexa platform) of high stiffness (aluminum plate) (Fig. 7) and that of low stiffness (PVC plate) (Fig. 8). In both cases the settling times obtained are approximately identical. Hence, owing to a mechatronic artifice the manipulator still 'sees' the same environmental stiffness irrespective of whether the contacted environment exhibits high or low stiffness. The minimal stiffness the manipulator has to do with is simulated by means of the Hexa-controlled stiffness.

The application of a controlled stiffness resulted in making the force control process as a whole insensitive to the variable environmental stiffness, and thereby in obtaining constant settling times independently of the stiffness of the material, with which the manipulator comes into contact. The additional advantage of the proposed solution is the ease of defining the environmental stiffness that can vary even during a single operation beat. The most obvious disadvantage of the proposed technique is its mechatronic complexity, particularly in the six-DOF case. However, the system under discussion should be considered as a solution being competitive with respect to complicated and expensive adaptive systems, or impedance control systems using passive flexibilities.

To be continued...
If you are interested on this project, please contact Rafal Osypiuk.

Peg-in-hole task using stiffness actuator.
Video download: PegInHole.wmv (Windows Media, 3.4MB)

It took 0.3s to generate this page.