Seite - 222 - in Proceedings - OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“

Table 1. Startand end positionof therobot startposition final position startvelocity final velocity θ1 0 ◦ −90◦ 0rad/s 0rad/s θ2 0 ◦ −10◦ 0rad/s 0rad/s θ3 0 ◦ 45◦ 0rad/s 0rad/s xTCP −0.15320m 0.81441m 0m/s 0m/s yTCP 0.92112m −0.15320m 0m/s 0m/s zTCP 0.02032m 0.22233m 0m/s 0m/s whichcontains thequadratical signalenergy tobeminimized. Thescrap-functionS ofEquation(14) describes theendpoint error and is specified by S(x,t)=α { β [y(x)−y]2+ [ ∂y ∂q q˙− y˙ ]2} (15) whereαandβareproperweightingfactorsandy, y˙containsthepositionandvelocityof theendpoint incoordinatesof the systemoutput. 5.2. Results The identification process of the signal energy optimal manipulation was started with the standard motion which is given from the robot controller. The results were verified on a real six-axis robot at thehome institution. Hence, thedata of the experimentand thesimulationresults are summarized in Figure 3. On the vertical axis the signal energy consumption is plotted over the time. It can be seen, that the standard manipulation wastes a lot of energy at the beginning and at the end of the motion due to the abrupt acceleration of the bodies. However, the signal energy optimal manipulation starts with a smooth movement of the heavy bodies. Therefore, the maximal speed of the axis have to be higher in comparison to the standard manipulation to reach the endpoint in the same period of time. Asaresult the reductionof thesignalenergyafter theoptimizationprocess isabout47%withrespect 0 10.2 0.4 0.6 0.80.1 0.3 0.5 0.7 0.9 0 50 100 Figure3. Build-up of the mechanicalenergyconsumption 222