Compensation of kinematic transmission errors on industrial robots (HyComp II)

工业机器人（IR）在更大的工作空间、更高的灵活性和更低的投资成本方面比传统机床具有显着的优势。然而，与传统机床相比，IR的较低的路径精度限制了其应用的可能性。特别是对于加工任务，其中路径精度是制造过程质量的决定性因素，IR的精度比传统机床差100至1000倍。这种情况阻碍了IR在许多加工过程中的广泛使用。出现路径误差的原因尤其归因于机器人关节或其传动系的非线性动力学。这些又是由所采用的精密传动装置及其由滞后引起的传动误差以及运动传动误差引起的。在第一个资助期内，对机器人传动装置的滞后和摩擦行为的多维性（负载、温度、频率）进行了研究和建模。在此基础上，提出了一种补偿迟滞引起的扭转的方法。在单轴测试台架上，证明与无补偿的基于电机的P-PI控制相比，这将跟踪误差降低了高达90%。剩余的误差部分主要是由于传动装置的运动传动误差。然而，这些对于高级加工任务来说仍然太大。第二个资助期的主要重点是调查和建模的多维运动传动误差及其补偿。这将使IR无需额外的传感器即可完成苛刻的加工任务。

Industrial robots (IR) offer significant advantages over conventional machine tools in terms of a larger work space, higher flexibility and lower investment cost. However, the lower path accuracy of IR compared to conventional machine tools limits their application possibilities. Especially for machining tasks, where the path accuracy is a decisive factor for the quality of the manufacturing process, the accuracy of IR is inferior to conventional machine tools by a factor of 100 to 1000. This circumstance prevents the widespread use of IR for many machining processes. The reasons for the occurring path errors are to be attributed in particular to the nonlinear dynamics of the robot joints or their drive train. These are in turn caused by the employed precision transmissions and their transmission errors resulting from hysteresis as well as kinematic transmission errors. In the first funding period, the multidimensionality (load, temperature, frequency) of the hysteresis and friction behavior of robot transmissions was investigated and modeled. Based on this, a method was developed to compensate for the torsion caused by hysteresis. On a single-axis test bench, it was demonstrated that this reduces the tracking error by up to 90 % with respect to the motor-based P-PI control without compensation. The remaining error portion is largely due to the kinematic transmission errors of the transmissions. However, these are still too large for advanced machining tasks. The main focus of the second funding period is the investigation and modeling of the multidimensionality of the kinematic transmission errors as well as their compensation. This should qualify IR without additional sensors to perform demanding machining tasks.