Active Child Models for Traffic Safety Research Interim Report 2, October 2013
The project Active Child Models for Traffic Safety Research is funded by Folksams Forskningsstiftelse. The overall project aim is to increase the safety of child car occupants and thereby reduce the number of traffic induced injuries in 3 to 12 year-old children. The specific aim is to create a computer model of a child that includes active musculature. Based on literature review of child numerical models it was decided to proceed with child multi body models in the MADYMO code (TASS, Rijswijk, the Netherlands). The 6 and 10 year-old child facet models, the Q6, Q10 and Hybrid III 6 year-old ATD models were compared regarding kinematics to experimental data with child volunteers in 1 g braking and steering events. The ATD models did not represent the experimental ATD response. The child facet models represented the child volunteers for about 3-400 ms of the events, and after that behaved cadaver like with much larger head and sternum displacement compared to the volunteers.
Then, the child facet model representing the 6 year-old child was chosen to implement muscle activity. Muscle activity was represented by an active spine that applied torques at each vertebral joint in response to joint angle changes in two directions: flexion-extension motions and lateral bending. A partial, integrative and derivative controller governed with input from angular sensors controlled torque actuators. The controller gains were based on adult data and scaled by 50% for a first version of the active child model. Then, an optimization approach was adopted to tune the control gains in the lumbar, thoracic and cervical spine such that head and sternum kinematics would correlate to the mean values from the volunteer braking and steering experiments. The first version of the active child model had a significantly improved biofidelity compared to the original facet model, with shape and magnitude of head and sternum displacements similar to the volunteer data. The optimization created non biofidelic gain combinations, although providing important input to future work. It is necessary to perform a reanalyses of the experimental data in order to have data for powerful performance criteria in future optimizations.
Future work is needed to improve the optimization and provide controller gains based on child volunteer data, rather than scaled adult data. Then, the tuned active child model has to be validated to new experimental data sets. A new steering and braking experiment with approximately 1 g acceleration loading was performed during 2013 and will provide a good validation data set. Also, there is a data set with child volunteers in sled test of approximately 4 g that can be used. Then, the active model is suitable to perform parameters studies of how child restraints design, emergency manoeuver characteristics and child posture influences the safety of children in the rear seat.