In the world of industrial robotics, the need of improving the precision is increasing. Nowadays the maximum precision reached by industrial robots is in the order of the micron. Improving the precision is difficult, because environmental factors start to play an important role. In fact, thermal variations are the principal cause in a loss of accuracy. Furthermore, the manufacturing process can also cause the dilatation of the robot parts. Therefore, the aim of this work is to study, model and compensate those environmental effects in order to improve the robot final accuracy.
Specifically, this research concerns the development of a new general calibration procedure suitable for all high-precision robots and machine-tools. Calibration techniques used until now do not allow the robot use in industrial environment, because they does not keep in account environmental factors. Furthermore, forces generated during the manufacturing process can also deform the robot structure. The calibration strategy proposed in this work will keep in account those effects, besides caring about standard geometric error compensation. This study includes the development of methodologies, convenient tools, measurement systems, software and algorithms to realize the objective of dynamically compensate the errors originated from the previous issues.

The objective will be achieved by mastering all the error origins singularly to finally combine them in a global dynamic compensation.

The following steps have to be accomplished:

  1. Development of an adequate measuring system, that includes:
    • Measure of translations and rotations at high-precision.
    • Measure of all the temperatures involved in the measuring loop.
    • Development of a contactless force simulator, to simulate manufacturing process forces on the end-effector.
  2. Considering the thermal drift, calibration of a 1 DOF linear axis.
  3. Considering thermal and manufacturing forces drift, calibration of a 3 DOFs parallel kinematic delta robot (Agietron Micro-nano).
  4. Considering thermal and manufacturing forces drift, calibration of a hybrid system composed of two 3 DOFs robots (Agietron Micro-nano and MinAngle).
  5. Calibration confirmation of the previous system by mean of indentation.
Parallelogram used in step 2. Agietron micro-nano, used during step 3.
The robot MinAngle. Agietron micro-nano and MinAngle, step4

Main achievements

The measuring system

The measuring system has successfully been developed: it is composed of two interferometers and two autocollimators. With this system it is possible to measure 5 DOFs (2 DOFs in translation and 3 DOFs in rotation) at the same time. The remaining translation is acquired in a second time. The interferometers are two SIOS SP-2000, with a resolution of ~1 nm and a stroke of 2 m, while the autocollimators are two Newport LDS-1000, with a resolution of 0.02 arcsec and a stroke of ±400 arcsec (±0.11°).
The temperatures of all the parts in the measuring loop are acquired using a total of 12 sensors: 10 pt1000 are used for parts of the measuring loop while 2 pt100 are used to measure the air temperature.
The measure of translations and rotations is strictly linked with the temperatures reading. For us is very important to divide the robot drift by the drift of the overall system. This has been achieved using special materials with an extremely low dilatation coefficient for the static components (like invar or zerodur) and by stabilizing actively all the supports in aluminum. Also the drift due to the air temperature drift is compensated electronically.
The force simulator is in phase of development and it will be ready by the end of 2008. This system is composed of 3 inductances, which are used to create a magnetic force on the robot end-effector. Such force has been dimensioned to be as much similar to the forces emitted during a standard EDM process.


The calibration of the 1 DOF linear axis has been achieved. The linear axis has an initial error in positioning superior to ±300 nm. After the geometric calibration the error passed to ±206 nm. Finally, including in the model also the prediction of the robot drift we reach the error of ±10 nm. The model has been built using the Stepwise regression algorithm of the software Matlab.
Thanks to this work it has been proved that the thermal calibration procedure is effective. The results are in the expected order of magnitude and are encouraging to continue on a more complex case.

Future achievements

The calibration of the 3 DOFs robot Agietron Micro-nano is in course. Once the force simulator will be ready, it will be mounted under the end-effector of the robot. In a first phase, it will be performed a thermal calibration of the robot. After it will be performed a “force calibration” in order to quantify the deformation due to the EDM process. Finally the robot will be calibrated keeping in account those two effects.

After this phase, the measuring system will be adapted to calibrate the robot MinAngle. This robot has been specially developed to work with the previously mentioned 3 DOFs robot. The aim of this part is to see how is it possible to mix two calibration processes: it will be proposed a strategy to unify two different calibration dataset, beyond a strategy to unify two different references.

Finally a calibration confirmation will be performed. The goal of this part is to confirm all the strategy by manufacturing a special piece and by measuring it. For this task, we propose the indentation process: a diamond will be mounted on the end-effector of the Agietron Micro-nano robot, while a substrate will be mounted on the robot MinAngle. Indents will be produced in the substrate at well-known distances. Thus, the indents distances will be measured with an electronic microscope. This validation will quantify the real precision of the two robots system in an industrial environment.