Study On Method Of Dynamic Parameter Identification Of Industrial Robots

The robotic technology is developing towards high speed, high accuracy and highintelligence, so it set a higher criterion for robot control algorithm. Compared to the PID controlalgorithm, which only take into account of the position and the velocity error of the robot, themodel based control scheme can inhance the dynamic performance and the track tracing accuracyof the robot by incoorperating the robot dynamic model. To construct the model based controlalgorithm, the robot dynamic model with high accuracy is needed. However, in the practical worldthere exist many non-ideal factors which can greatly affect the accuracy of the robot dynamicmodel due to the harsh working environment. To reach high control accuracy, these non-idealfactors must be compensated. Robot dynamic parameter identification is an ideal compensationmethod, which can compensate most of the dynamic affecting factors without increasing thecomplexity of the dynamic model. In this paper, two different dynamic parameter identificationmethods, the Linkage-assemblies based identification method and the Torque-currentmeasurement based identification method, are introduced according to different accuracy criterionin different application tasks. Comparing to the existing identification methods, theLinkage-assemblies based identification method can reduce the computation burden and thecumulative error by identifying all the dynamic parameters in several steps, and theTorque-current measurement based identification method can discouple the identification of theinertial parameter and the friction parameter by identifying inertial parameters using torque dataand friction parameter using current data, so can inhance the identification accuracy further more.Firstly, the paper construct the dynamic model of serial robot system, deduce the linear formof the dynamic model which is used in the identification process as the identification model,analyze in depth the dynamic property of the serial robots and the factors affect the accuracy of thedynamic model, elaborate the merits of dynamic parameter identification method in compensatingthese factors and the main parts of the identification process.The relationship between the type of joint motion and the identificablity of every parameter isanalyzed through the robot energy function. By this analysis method, one can deduce thenecessary joints need to be actuated in order to identify the aim parameter without symbolderivation of the dynamic model, which is very tedious and time-consuming especially for therobots with many joints. This analysis can be used as a guidline for the Linkage-assemblies basedidentification method to decide which joint can be locked and which joint must be actuatd in orderto identify the target parameters. The independce of the parameters of the same linkage is proved by the mathematical method based also on the robot’s energy function.The existing identification method is categorized into two types: Current-measure based andtorque-measure based method. Their respective identification principle and identification processare illuminated in detail, and their drawbacks are analysed too. Taking into account of the expenseand the accuracy need to be reached, we introduce the Linkage-assemblies based identificationmethod and the Torque-current measurment based identification method, which can overcome thedrawbacks of the existing methods to some extent. Their respective identification priciple,identification process, mathematical model and identification algorithm is illustrated in detail. Thesimulation of these two identification method is elaborated too. The comparision of the simulationresult with the result acquired by the existing methods testify the fesiblity of our method.