Blue box: a device to rebuild the 3D kinematics and the actions on the drivers along a road path

A new instrument, named blue-box, has been designed and realised to rebuild the complete 3D kinematics of a rigid body, including linear and angular components of velocity and acceleration. The device is an electronic unit designed to house the sensing elements, i.e four triaxial MEMS accelerometers (by STMicroelectronics) placed inside the box in an optimised configuration. The errors produced by inaccurate sensor placements and orientations, or by the non-linearity of single sensing components are corrected by preliminary calibration. The calibration is composed by two main steps: a static procedure firstly corrects the offsets and gains spreads of each accelerometer placed inside the blue-box, and secondly adjusts the errors caused by imprecise orientation of sensor frames. The next step consists in a dynamic procedure, where the box is mounted on a rotating platform, whose angular velocity and acceleration are feedback controlled. Measuring the sensors outputs during rotation, it is possible to reconstruct the blue-box angular velocities and accelerations by the solution of an over determinate linear system according to the pseudo-inverse matrix solution method. The results of the dynamic calibration procedure permit to correct also the errors related to the inexact sensor frames positions. After complete calibration, the blue-box is ready for outdoor trials to acquire the data required to perform kinematics reconstruction of the rigid body to which it has been fixed. In spite of its versatility, the blue-box has been specifically designed to compute and acquire the kinematics of road vehicles and, by means of a mathematical model of the human body, the actions transmitted to the car occupants. In particular, it will be used for the calibration of a new car driving simulator, which is under development by our research group; in this case the most important parameters to be evaluated are the linear and angular accelerations which are required to compute the inertia actions acting on the subject. The present paper deals with the calibration procedure hints and describes the transformations to be applied to sensors raw data to obtain the minimisation of the reconstructed kinematics variables errors.