In EU-funded BALANCE project, developing a stability index which can be employed to estimate actual state of balance of both healthy and neurologically impaired humans' walking in exoskeleton was one of scientific tasks. In the task, Centroidal Momentum (CM), referring to linear and angular momenta at Center of Mass (CoM), has raised as a potential index for such purpose. While our past studies have presented analysis results of CM in offline and online (real time) manners for walking of healthy human and stroke patients, in this study, we present real time computation of CM in exoskeleton-supported walking, specifically with healthy subjects. Experimental setup consists of LOPES II, a treadmill-based robotic gait training exoskeleton for lower limbs rehabilitation developed by Twente University, and commercially available IMUs (Inertial Measurement Units)-based full body motion capture suits from Xsens. CM was computed and demonstrated in two walking conditions: unperturbed walking and walking with unexpected pelvic perturbations in lateral direction. While electromagnetic fields (EMF) from LOPES II exoskeleton affected signals of IMUs in the motion capture suit, the results show the potential applicability of the CM as a sort of stability index for human walking in the exoskeleton.