Calibrating force/torque (F/T) sensor drift is an enduring objective for robotic precise force control. This article presents a novel drift identification method to discover the dynamics of F/T sensor drifts from nois...
详细信息
Calibrating force/torque (F/T) sensor drift is an enduring objective for robotic precise force control. This article presents a novel drift identification method to discover the dynamics of F/T sensor drifts from noisy measurement data, which is conducive to accurate sensor drift compensation. In the drift identification method, a linear dynamical model with measurement noise is formulated to characterize the evolution of sensor drift, and an expectation-maximization optimization framework which integrates Kalman smoothing with sparse Bayesian learning is put forward to identify the parameters of the linear dynamical model using F/T sensor measurement data. The effectiveness of the proposed drift identification method is validated on extensive robotic experiments including scenarios with unloaded mass, loaded mass, and contact force. Experimental results demonstrate the superiority of the proposed drift identification method compared to the conventional least square method for sensor calibration.
Vocal tract length normalization (VTLN) has been successfully used in automatic speech recognition for improved performance. The same technique can be implemented in statistical parametric speech synthesis for rapid s...
详细信息
Vocal tract length normalization (VTLN) has been successfully used in automatic speech recognition for improved performance. The same technique can be implemented in statistical parametric speech synthesis for rapid speaker adaptation during synthesis. This paper presents an efficient implementation of VTLN using expectationmaximization and addresses the key challenges faced in implementing VTLN for synthesis. Jacobian normalization, high-dimensionality features and truncation of the transformation matrix are a few challenges presented with the appropriate solutions. Detailed evaluations are performed to estimate the most suitable technique for using VTLN in speech synthesis. Evaluating VTLN in the framework of speech synthesis is also not an easy task since the technique does not work equally well for all speakers. Speakers have been selected based on different objective and subjective criteria to demonstrate the difference between systems. The best method for implementing VTLN is confirmed to be use of the lower order features for estimating warping factors.
暂无评论