Distributed robust vehicle state estimation

Abstract

Abstract-A distributed estimation approach based on opinion dynamics is proposed to enhance the reliability of vehicle corners' velocity estimates, which are obtained by an unscented Kalman filter. The corners' estimates from a Kalman observer, which is formed by integrating the model-based and kinematic-based velocity estimation approaches, are utilized as opinions with different levels of confidence in the developed algorithm. More reliable estimates robust to disturbances and time delay are achieved via solving a convex optimization problem. Road tests confirm the robustness of the methods independent of the powertrain configuration on surfaces with various friction conditions in pure and combined-slip maneuvers, which are arduous for the current vehicle state estimators.

Linear TDOA-based Measurements for Distributed Estimation and Localized Tracking

“…Algorithm 1: distributed κ-redundant estimation and localized FDI. 1 Input: matrices F, G (or sampling period T ), FAR κ, redundancy factor κ 2 Design the networks G (the sensor network) to be SC and κ-connected, e.g., via algorithms in [33], [34]; 3 Calculate the TDOA measurement matrix H i via (14) based on the relative positions of sensors; 4 Design the block-diagonal gain matrix K via LMI in [16,Appendix]; 5 begin sensor i = 1 : N at every time k Finds the estimate x i k|k via Eq. ( 15)-( 16) on the target position (state) via the received information; 9 Finds the residual r i k or distance measure z i k ;…”

“…1]. These distributed strategies are mainly developed over linear and partially observable (measurement/system) models, e.g., for model-based and kinematic-based vehicle speed estimation [14]. In this direction, the nonlinear rangebased methods as TDOA [4], [5], [15], are linearized to be adapted for distributed setups.…”