<p>All mechanical systems with moving parts are affected by friction, including industrial robots. Being able to design an accurate friction model would further increase the performance of todays robots. Friction is a complex dynamic phenomena that is constantly changing depending on the state and environment of the robot. It is therefore beneficial to update the parameters of the friction model online. An estimate of the friction will be made using the feedback control signal with the help of a feedforward control scheme in a two axis simulation setup. The friction estimate is then used for an offline identification of three friction model parameters in a static Lugre friction model. Improvements on the identification will be done by introducing some shut-off rules that will improve the estimate. The normalized least mean square method (NLMS) will then be used to update the parameters online. A simulation of friction compensation with a fixed friction model, and with an adaptive friction model will be studied. The method will also be simulated using experimental data taken from a real industrial robot.</p>
| Identifer | oai:union.ndltd.org:UPSALLA/oai:DiVA.org:liu-19269 |
| Date | January 2009 |
| Creators | Längkvist, Martin |
| Publisher | Linköping University, Department of Electrical Engineering |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, text |
Page generated in 0.0018 seconds