yes it is lil hard, you need to find: angle of inclination (foot contact with floor), BW, acceleration during heel strike, and to add Ms force PF and QL, let me find it and get back to you

The short answer is that it is not possible to calculate GRF/GRMs with joint kinematics and without the use of force sensors under the subject's feet. There is research on the prediction of GRF/GRM with the use of kinematics and machine learning, but those models need to be trained first, they apply only to specific conditions (e.g., treadmill running), and you can't really find a ready to use solution for your problem. However, since you are using 3d mocap, if you are able to estimate the kinematics of the person's centre of mass (or approximate with a single marker in the area of the posterior mid pelvis), you may multiply the acceleration of the CoM with the mass of the subject and that would return an approximation of the total GRF. However, this approach is a) inaccurate since you practically measure skin acceleration and not CoM acceleration, and b) there is no way to accurately distribute the GRFs between the two legs during bipedal foot contact. However, this approach may work in single leg stance or in running. Theoretically, if you have the GRFs, you may then use inverse rigid body dynamics, joint angles and anthropometric measurements to calculate GRMs around a joint. However, even in that case, you will encounter difficulties since the centre of pressure during the recorded activity is not a given, so it will need to be approximated.

Sokratis Dimitrios Komaris - thanks so much for your reply. Very insightful.

Someone else has informed me that it is possible using the software AnyBody and/or Opensim MOCO. Would you advise against such an approach? I am starting to lean towards sticking to predictive simulations to estimate motion & resultant GRFs, rather than using my old dataset of 3D mocap. Thanks for your comments!

Ashleigh Wiseman you are absolutely right, the part about a ready-to-use solution on my previous answer was misleading. AnyBody can estimate GRFs from MoCap data (I don't really know about MOCO) in a similar fashion to what i described. It really depends though on what type of activity you are recording and the level of accuracy you need in your predictions. I also expect that full bod mocap would return more accurate results.

When only one foot is on the floor, it is easy. Are you working on walking or running? Take anthropometric tables. Computes the displacement of the center of mass of the whole body from the movements of the body segments. Derive twice to have the acceleration. Then F=m*a. However the result is based on assumptions on the mass of different body segments and is impacted by soft tissue artefact and other kinematic data errors. You can find some papers on that. For the double support phase of walking it is much more complicated.