PAMM · Proc. Appl. Math. Mech. 17, 201 – 202 (2017) / DOI 10.1002/pamm.201710070
A movement generation algorithm for FE Human Body Models
, Syn Schmitt
, Alexandra Bayer
, Julian Blaschke
, and Christian Mayer
Institute of Sports and Movement Science, University of Stuttgart, Allmandring 28F, 70569 Stuttgart, Germany
Stuttgart Research Centre for Simulation Technology (SimTech), University of Stuttgart, Germany
Daimler AG RnD Center, 70546 Stuttgart, Germany
Finite element (FE) method simulations are increasingly used for the development in the area of vehicle safety nowadays.
Highly detailed virtual mechanical and human body models (HBMs) available for use in connection with the increase of the
processors performance and algorithms efﬁciency, give engineers the opportunity to simulate not only the car crash event itself
but also a so-called pre-crash phase. This is important for the design and improvement of steering-assist and autonomous
driving systems, through the assessment of active occupant behaviour during the impact avoidance or any other complex
driving manoeuvres. To enable adequate evaluation of such simulations, virtual Active Human Body Models (AHBMs)
should be established, capable to not only reproduce reﬂex human reactions but also for simulations of human movements.
This study investigates the applicability of a forward dynamics movement generation algorithm for the FE HBMs, presents
ﬁrst results and outlines questions, need to be solved in future to do such simulations in a robust and time effective way.
2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
One aim of the automotive industry in the near future is an autonomous self-driving car with free sitting posture for the
occupants . He or she will be allowed to do a lot of other movements during “driving” in a car in addition to nowadays
steering. This situation directly affects the car’s active and passive safety systems elaboration and improvement, also taking
into account a steady increase in the road trafﬁc worldwide. New products are developed involving FE computer simulations,
and technical progress makes possible to cover more and more demanding application ﬁelds each year. Presently, the main
focus is the in-crash scenario and it will move more to pre-crash in the future. There, re-active and pro-active response of the
occupants to the road situation before possible impact or manoeuvre should be taken into account.
While back in the 1980s, FE models with few thousands of elements were used for the crash events simulations, modern
models can contain several millions of them. The engineering of very precise and detailed models has led to the fact, that the
human occupant models were included in the crash simulation and to the safety systems development process. Those systems
interventions are employed in a way to minimize the risk of occupants injury, depending on their kinematics and position.
And the injury risks itself could be determined from the FE HBMs using stresses or accelerations in post processing.
Originally, FE HBMs were created using cadaver material properties, which led to overestimated tissue stiffness, with
highly simpliﬁed muscle modelling, each of which consists only of a passive spring and damper element, implied the im-
possibility of movement simulations. As occupant position may be highly varying and individual depending on the muscles’
activation before and in a crash  and primarily, the muscular apparatus accounted for the active human movement, it could
not be neglected any more both in the crash and in the pre-crash phase. Thus, there is a clear need to implement general
movement generation algorithms for virtual FE HBMs, which are used currently for in-crash simulations, with the intention
to extend their application ﬁeld to pre-crash simulations.
2 Forward dynamics movement generation algorithm for FE HBMs
Based on a recently published forward dynamics full human body model simulation driven by 260 Hill-type muscle-tendon
units (MTUs) , it is proposed here to utilize the same approach for movement generation within the theory of ﬁnite elasticity
and ﬁnite element method.
There are two main approaches to human movement control: one is based on the joints torques control  and second
- on the internal control of calculated a priory muscle stimulations . Both methods require an inverse problem solution,
resulting in a high simulation complexity and great sensitivity to disturbances. Thus both methods are hardly applicable in the
perspective of a FE model applications with millions of degrees of freedom (DOF). Equilibrium point (EP) control is proposed
to be used instead. The main idea of this approach is that so-called EPs are to be reached at different instants in time, in which
both joint positions as well as individual muscle lengths are known and therefore can be used as a target measure (reference
variable) for the control. This allows for calculating corresponding muscle forces, activation and stimulation during runtime in
a more cost-effective manner, depending on the deviation of the actual position from the EP position . One of the variants
of EP control is the λ-control , which is based on the control of the length of each individual muscle, assuming that desired
Corresponding author: e-mail firstname.lastname@example.org, phone +49 711 685 68040, fax +49 711 685 50484
2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim