Design and implementation of an in-vehicle human state estimator: real-time human kinematics estimation based on an active multibody human model

This report presents a project about the feasibility study of developing a real-time in-vehicle human state estimator that identify occupant kinematics in pre-crash scenarios. The report starts by stating the background on road safety and the motivation of developing the estimator.
It is predicted by simulation with a commercial modeling package that the occupant pose before a crash has a significant influence on injury level as a result of the crash. The injury level can be reduced by 35 % by optimizing the occupant posture. The development of the estimator follows an iterative system engineering approach using V-model. A list of stakeholder was identified, and customer requirements were collected. After that, the conceptual design was performed; the design includes system analysis, scenarios, system views, and requirements analysis. Then a human model is adapted to present an adult male driver. The estimator, which is intended to be applied in pre-crash braking, is designed in Simulink and implemented in a dSPACE setup. The estimator design is verified in a virtual environment and a real life environment. The report concludes by mentioning conclusions, recommendations, and the project management activities of the project.