Content area
Abstract
This thesis proposes a new approach to the dynamic parameter identification of robotic manipulators that overcomes some of the difficulties usually encountered with standard methods and allows better identification results. In this approach, dynamic parameters are sequentially identified in three steps, which can avoid the use of a large amount of data simultaneously and resulting high computation complexity. The joint friction models are initially obtained, then the gravity parameters, and finally, all the other parameters.
Moreover, inertial measurement units (IMUs) are suggested in the identification process and expected to allow for a more precise estimation of the joint velocities and accelerations when compared to the conventional technique, namely the numerical differentiation. The robust design scheme for an IMU composed of only accelerometers is developed. The robustness to the uncertainty of the locations of the sensors and the measurement noise is obtained through redundancy and optimal configuration of the onboard sensors. The fail-diagnostics and fail-safe issues are also addressed for the reliable operation of the unit.
Furthermore, two alternatives for friction modeling are proposed to account for joint friction in order to cope with its highly non-linear characteristics, which cannot be properly handled by a simple model in practice. The first approach is using a polynomial function in terms of the joint positions and velocities, while the other employs fuzzy logic.
The thesis also investigates the issue of joint flexibility, which is particularly challenging in the dynamic identification as well as the control of robotic manipulators. A method to estimate the flexibility parameters of non-rigid joints is proposed. Theoretical results are applied to identify the joint damping and stiffness parameters with a planar one- and two-flexible-joint manipulator in computer simulations.
