Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Using Active Learning Methods for Human-Seat Interactions To Generate Realistic Occupant Motion

Niklas Fahse, Monika Harant, Marius Obentheuer, Joachim Linn, Jörg Fehr

Abstract

In the context of developing new vehicle concepts, especially autonomous vehicles with novel seating arrangements and occupant activities, predicting occupant motion can be a tool for ensuring safety and comfort. In this study, a data-driven surrogate contact model integrated into an optimal control framework to predict human occupant behavior during driving maneuvers is presented. High-fidelity finite element simulations are utilized to generate a dataset of interaction forces and moments for various human body configurations and velocities. To automate the generation of training data, an active learning approach is introduced, which iteratively queries the high-fidelity finite element simulation for an additional dataset. The feasibility and effectiveness of the proposed method are demonstrated through a case study of a head interaction with an automotive headrest, showing promising results in accurately replicating contact forces and moments while reducing manual effort.

Towards Using Active Learning Methods for Human-Seat Interactions To Generate Realistic Occupant Motion

Abstract

In the context of developing new vehicle concepts, especially autonomous vehicles with novel seating arrangements and occupant activities, predicting occupant motion can be a tool for ensuring safety and comfort. In this study, a data-driven surrogate contact model integrated into an optimal control framework to predict human occupant behavior during driving maneuvers is presented. High-fidelity finite element simulations are utilized to generate a dataset of interaction forces and moments for various human body configurations and velocities. To automate the generation of training data, an active learning approach is introduced, which iteratively queries the high-fidelity finite element simulation for an additional dataset. The feasibility and effectiveness of the proposed method are demonstrated through a case study of a head interaction with an automotive headrest, showing promising results in accurately replicating contact forces and moments while reducing manual effort.
Paper Structure (11 sections, 8 equations, 7 figures)

This paper contains 11 sections, 8 equations, 7 figures.

Table of Contents

  1. Introduction
  2. Methods
  3. Human Behavior Prediction with an Optimally Controlled Biomechanical Multibody Model
  4. Surrogate Model of Occupant-Seat Interaction
  5. An Active Learning Approach for Simultaneous Training Data Generation and ML Model Training
  6. Modeling and Results
  7. Multibody Model
  8. Finite Element Model
  9. Multilayer Perceptron
  10. Active Learning Loop
  11. Conclusion and Further Work

Figures (7)

  • Figure 1: Active learning approach for generating training data and training the ML model.
  • Figure 2: Kinematic structure of the MBS model with one degree of freedom.
  • Figure 3: Multibody simulation of the head movement toward the headrest. (a) Phases of the simulation with varying cost functions. (b) Joint angle and (c) joint actuation of an exemplary solution.
  • Figure 4: Reduced FE model of the head and the headrest. The headrest consists of a cushion and a headrest bar. The head is reduced to the parietal and occipital bones and the covering skin.
  • Figure 5: Exemplary FE simulation of head movement toward the headrest. In (a), snapshots at times $t = 0\, \textrm{s}$ and $t = 1\, \textrm{s}$ are shown. In (b), the relevant coordinates of the interaction forces and moments acting on the head, given in the headrest coordinate system, are plotted. Initially, the head is separated from the headrest. First contact occurs at $t = 0.29\, \textrm{s}$.
  • ...and 2 more figures