State estimation is the basis for condition monitoring and advanced control of industrial robots. To accurately estimate the system state, additional sensors are required to capture the motion of industrial robots. A common solution for robot motion capture is to install inertial sensors on robots and calibrate their poses (position and orientation). Existing methods for sensor pose identification consist of two steps. The first step is to estimate the sensor orientation and the second step is to estimate the sensor position. Therefore, these methods are subject to the error propagation problem. To tackle this problem, this paper proposes a new method for the pose identification of inertial sensors. In the proposed method, the nonlinear mapping between the joint state variables and the inertial measurements is first established based on the spatial vector and the product of exponentials formula. Next, the identifiable condition is constructed and the finite Fourier series trajectory with a specific bandwidth is adopted to excite the robot to satisfy the identifiable condition. Finally, the pose of the inertial sensor is obtained by solving nonlinear equations. The proposed method is able to avoid the propagation of errors by simultaneously identifying the orientation and position of sensors. The effectiveness of the proposed method is verified by simulations on a serial robot with six degrees of freedom.

Parameter identification; Kinematics; Inertial sensors; Industrial robots