A Digital Image Correlation technique is proposed to control a quasi-static 6 degrees of freedom testing machine. This machine is based on a hexapod architecture, allowing displacements of several tens of centimeters and degrees, and with force capacities of several tens of kN and kN.m. The control technique is based on the measurement of the set of actuator lengths, from images of the end-effector of the machine. A firstattempt is shown with a single camera. Thoughrelevant, the results present too high uncertainties for the aimed applications. Several cameras are then used to improve the 3D rigid body motion displacement measurement of the end-effector. Thecamera positions are free, with an automatic calibration method assessing the needed coefficients for the control. The set of actuator lengths is then searched by global minimization of the difference between the current image and the reference image of each camera (i.e., the minimization is not performed independently for each camera). The algorithm is implemented on Graphical Processing Units to achieve computation times lower than 50 ms. An in-depth experimental validation of the whole setup is performed. For an actuator length range of 200 μm, the actuator length uncertainties are around 0.4 μm with 3 cameras. The errors of displacement of the end-effector are less than 2 μm, partially due to the validation setup itself.

A new experimental setup for concrete mixed-mode crack propagation is presented. The specimen and loading correspond to a Nooru-mohamed test, i.e. a square double-notched specimen submitted to in-plane tension and shear. Innovation comes from the loading setup: a hexapod machine whose 3D motions are controlled by a Digital Images Correlation algorithms using several cameras “live” recordings. The interest of this setup is that the 3D (in-plane and out-of-plane) boundary conditions will be controlled contrary to the original Nooru-Mohamed setup, which conditions have a tremendous influence on the crack propagation, and thus further confrontations to numerical simulations.