Call2 Monitoring ODEUO

Public Summary Month 7/2012

Mon, 23 Jul 2012 16:49:05 GMT

The ODEUO experiment was concluded using more vibrating masses at the same time and finally getting observations of “real” test objects.

Different mounting suspensions coming from the automotive industry as well as packaging boxes were evaluated.

Example of a mesurement cycle to detect damages on packaging boxes

For dissemination purposes the experiment is made public on the homepage of the robotic institute of the Linz University, and linked to echord homepage (see www.robotik.jku.at – research – echord).

We also participated at the conference OPTIROB 2012 in Romania, Mamaia which was held from the 21 to the 23 of June. (http://www.wix.com/olaruadrian/optirob2012)

The paper “A Pneumatically Driven Stewart Platform Used as Fault Detection Device” is printed in the journal of the conference and attached here (OPTIROB_Hexapod_LaTex.pdf), the conference presentation is available here (OPTIROB2012_ODEUO.pdf).

This picture shows the basic setup of the ODEUO experiment, for sensorless detection of unknown vibrating objects.

See also some videos of the test bench (testbench.mp4) as well as tests with “real” objects like motor suspensions (motor_suspension.mp4) or packaging boxes (packaging_boxes.mp4).

A multimedia report of the whole experiment is visible here, in different video formats ( Echord_Odeuo_Multimedia_5_640x480.wmv, Echord_Odeuo_Multimedia_5_640x480.mp4, Echord_Odeuo_Multimedia_5_640x480.mov, Echord_Odeuo_Multimedia_5_640x480.avi ).

Video for mobiles in low resolution: Echord_Odeuo_Multimedia_5_640x480.3gp

Public Summary Month 16/2012

Thu, 31 May 2012 14:21:18 GMT

The experiment with different vibrating masses vibrating into different directions showed a very good detection sensitivity of low vibrating masses, down to 0,5% of the payload range. This means that with the Stewart platform capable to move 20kg, a mass of 100g had been detected, if it was vibrating in the predefined frequency range between 5 and 18 Hz.

Moreover the verification of the theoretical approach gave an excellent accordance between model and physical setup, in the first three of the six degrees of freedom that were analyzed.

In the final project stage this verification will be extended to all six DOF’s of the hexapod.

The experiment is made public on the homepage of the robotic institute of the Linz University, and linked to echord homepage (see www.robotik.jku.at – research – echord).

We also participate at the conference OPTIROB 2012 in Romania, Mamaia which is held from the 21 to the 23 of June. (http://www.wix.com/olaruadrian/optirob2012)

The paper title is “A Pneumatically Driven Stewart Platform Used as Fault Detection Device” and is printed in the journal of the conference.

Attached you can find a video of the actual test bench ( Echord_Odeuo_640x480.mp4)

Public Summary Month 14/2012

Fri, 30 Mar 2012 09:16:02 GMT

The implementation and verification of the experiment continues, using the special designed spring and mass system.

In this way the vibrating mass can be fixed to the hexapod in different angles and at different positions.

Finally at end of experiment different vibrating masses, positioned at different positions should be identified.

Starting with the results of the last 2 months, vibrations were clearly identified using a mass of 500 g and shaken in vertical direction.

In the next months measurements will be spread to more different masses vibrating in different orientations.

Dissemination

There was a video presentation for the "eu robotics week" http://www.robotics-labs.eu see Institute for Robotics - Linz University.

The ODEOU project will be pubilshed at the homepage of the Institute within the next weeks.

We plan a poster presentation at the automatica fair in munich in may 2012 and

a conference participation at OPTIROB 2012, Mamaia, Romania.

Public Summary Month 1/2012

Fri, 27 Jan 2012 12:00:47 GMT

The theoretical base to identify the transfer functions of the vibrating hexapod with and without lose mass was elaborated and is described in milestone 2 (M12)

The experimental implementation and verification of this theory was started in the last month, using vibration sweeps of the hexapod from 0 to 30 Hz.

In the attached movies this movements are visible, first in slow motion, to distinguish all 6 degrees of freedom, and also in real speed vibrating in the vertical direction with a frequency going from 0 to 30 Hz.
(see odeuo_slow.wmv and odeuo_sweep.wmv)

In the next steps these measurements will be repeated with a mounted vibrating test mass, trying to identify it sensorless.

Public Summary Month 10/2011

Tue, 29 Nov 2011 08:46:00 GMT

Parameter identification

The frequency model described in month 8 works for a one dimensional excitation with a centered mass.

This theoretical result has to be prooved in a first experiment.To verify it on the moving platform a special test mass was developed. The basic test assumption states that the load separates in two masses connected via a spring. A flat spring connecting base mass and additional vibrating mass was used, to have a defined direction of movement.

Hardware adaptation

Adaptation of experimental hardware was started with focus on a new electronic control hardware, to overcome synchronization errors in the control of more DOF’s.

The hexapod is drived by 6 pneumatic actuators, each of them has an integrated position sensor, a pressure sensor, a servo valve and/or a digital valve. These signals have to be synchronized through all degrees of freedom.

The design was therefore chosen following a modular structure.

Public Summary Month 08/2011

Thu, 22 Sep 2011 12:24:43 GMT

The hexapod vibrates with different frequencies while a loose part is placed on its top.

Vibrations are shown in the two attached videos. One with a fixed part, the other with

a vibrating loose part.

First steps to identify the loose part are done.

http://www.echord.info/file/Attachments/blogs/odeuo/public-summary-month-08-2011/Hexapod_anim_6_B_fixed.mp4

http://www.echord.info/file/Attachments/blogs/odeuo/public-summary-month-08-2011/Hexapod_anim_6_B_loose.mp4

Public Summary Month 6/2011

Tue, 16 Aug 2011 14:12:27 GMT

Experiment description

Countless everyday life products such as electronic equipment or complex parts of the automobile industry suffer from physical or mechanical strain like vibration and shock.

The product design has to take all these stress factors into account to guarantee the full function of a product during shipping and final usage.

It is indispensable to test and measure all these stress factors beforehand. Usually those sorts of vibration tests are carried out on special vibrating units or shaking platforms.

Electric shakers are being used mainly for higher vibration frequencies in one dimension.

To test for more complex movement hydraulic multi axis motion units are commonly used, each custom build for a specific vibration pattern.

To overcome this drawback one could use a hexapod instead, which is capable of complex movements in all 6 degrees of freedom and is easy to reconfigure.

If a different pattern is needed one has only to change the trajectory of the hexapod.

In general a hexapod consists of a base and a plate connected through 6 actuators each independent of each other and capable of push and pull forces.

The latest realization at JKU uses pneumatic muscles which are only capable of contraction force thus it wouldn’t be possible to apply a load.

To compensate this a centered spring for expansion force is integrated in the design of the muscle hexapod.

The spring is highly nonlinear and not well specified. But the given design is light weight, easy to transport and environmentally safe,

no oil or chemical spill like hydraulic systems.

The aim of the project is to determine if a test load put on the hexapod changes in behavior i.e. to sense if something gets loose or even breaks.

In Task 1 the dynamical model has to be elaborated and a simulation of the hexapod computed.

The computation of the equation of motion is done in Maple and integrated via a dynamic model package (Simcode2) to Matlab Simulink as a C-code S-function.

A virtual reality model is used to visualize the simulation results as well as various scopes to investigate the model behavior.

It is possible to input the actuator force into the model together with a disturbance force and torque stated in the inertial system.

The output of the simulation is the actuator position together with the velocity.

The simulation result shown in this figure displays the output of a sinusoidal phase shifted actuator force input.

The next step is to investigate the behavior with changing connection parameters for body P and B and how it will affect the simulation results.

This model can now be used for further investigation on control and observer structures.

State month 6

A complex dynamical model is derived with the opportunity to simulate changing load distribution and load failure with different scenarios.

The simulation was implemented in Matlab Simulink and visualized by a vrml model to investigate the load model behavior.

Furthermore implementations to recognize load failure can now be tested safe, easy and fast.

The theoretical model of the hexapod can be used for a model based observer structure.