Session: 08-07-02: Novel Control of Dynamic System and Design II

Paper Number: 166124

Iterative Augmented Lead Measurement Method for Efficient Nonlinear Motion Estimation With Delayed Measurements 

This paper employs an iterative augmented lead measurement method to reduce the computational burden of nonlinear motion estimation with delayed measurements. The motion estimation in three dimensions consists of the estimation of sixteen states represented by a nonlinear state-space model. When measurements are delayed due to processing or communication time, the recent delayed measurements should correct the history of the estimation, significantly increasing the computational cost. In motions with high nonlinearities or slow update rates, the extended Kalman filter (EKF) may diverge, highlighting the importance of the unscented Kalman filter (UKF) and cubature Kalman filter (CKF). The augmented method converts the delayed state-space representation into a non-delayed one by increasing the order of the system, facilitating the use of EKF, UKF, and CKF. However, this increases the order of the Riccati equation and the risk of divergence in fast motions and inaccurate noise characteristics. The augmented lead measurement method employs the idea of iterative estimation in each update by reorganizing the measurements to align states. This technique can introduce a non-delayed system with the same order that can be employed by EKF, UKF, and CKF.

The proposed method utilizes augmented lead measurements to convert the delayed state-space into a non-delayed state-space with the same state dimension. Subsequently, iterative estimation techniques are developed for nonlinear estimators, including the extended Kalman filter, unscented Kalman filter, and cubature Kalman filter. These nonlinear estimators are used to handle the complexities of systems with highly nonlinear behavior, making them suitable for various real-world applications. The proposed method is compared with the conventional augmented method in one hundred simulations of a three-dimensional motion estimation scenario with random initial conditions. The simulation scenario consists of a stereo camera, lidar, and inertial measurement unit (IMU) to capture the motion of a rigid body through a kinematic model in three-dimensional space. Quaternion parameters are used to represent the orientation in the kinematic model. The stereo camera and lidar have delays due to the processing time required to provide the results, while the IMU is assumed to operate in real time with negligible time delay. The proposed method achieves considerably lower computational time and order of complexity than the augmented method. Moreover, unlike the augmented approach, the augmented lead measurement method preserves the positive definiteness of the covariance matrices and offers slight improvements in correcting initial guess errors. The results indicate the significant higher computational efficiency of the proposed method in comparison with augmented method, especially for UKF and CKF estimators. This makes the method highly applicable in scenarios that require real-time estimation with limited computational resources.

Presenting Author: Hamed Mozaffari SIUE

Presenting Author Biography: Hamed Mozaffari is currently a Ph.D. student at Southern Illinois University Edwardsville. He earned his first Ph.D. from K.N.T. University of Technology in Iran. In his first Ph.D., he focused on human driver models and their applications in advanced driver assistance systems and autonomous vehicles. He combined motivational psychological theory with the Lyapunov method and vehicle dynamic models to regulate human emotional responses in rapid near-collision scenarios. He identified the role of delayed measurements in rigid body estimation within his model, and in his second Ph.D., he is researching the estimation of the position and orientation of rigid bodies under delayed measurements to improve its estimation accuracy. His current research focuses on addressing the quaternion norm constraint through the augmented method or the quaternion manifold for delayed measurements.

Authors:

Hamed Mozaffari SIUE
Arman Dabiri INTELAG LLC,