Design of an adaptive human state estimator

  • G.J.A. Verhaeg

Student thesis: Master

Abstract

Active Human Models (AHM) for use in (pre-)crash analysis are important to understand the kinematic behaviour of a driver of a vehicle. Safety systems make crucial decisions in emergency or critical scenarios. Safety systems can benefit from information about the driver. The kinematics of the driver can be estimated using an optimal estimator, that combines a limited number of measurements with information from a realtime human model. In this paper, an AHM for a Kalman Human State Estimator (HSE) that uses one sensor measurement is adapted to accommodate for changes in attentiveness and neural delays of the driver. The model is adaptable via the controller, that represents the muscle activity of a human model. Different optimization criteria are considered for the estimator, including an H? optimal design. It is concluded that the use of one sensor leads to limited performance in the estimation, and that the use of multiple sensors is recommended.
Date of Award30 Apr 2014
Original languageEnglish
SupervisorL. van Rooij (External coach) & O.M.G.C. op den Camp (External coach)

Cite this

'