Int J Interact Des Manuf DOI 10.1007/s12008-011-0132-x ORIGINAL PAPER Biomechanically-based motion control for a digital human Giuseppe Di Gironimo · Luigi Pelliccia · Bruno Siciliano · Andrea Tarallo Received: 18 October 2011 / Accepted: 14 November 2011 © Springer-Verlag 2011 Abstract An algorithm for both human-like motion gen- eration and joint torques computation for digital humans is addressed in this paper. This goal has been achieved using techniques derived from robotics. In particular, the so-called augmented Jacobian has been used to solve the inverse kine- matics problem with a single closed loop inverse kinematics algorithm. Furthermore, a position control for the center of mass of the kinematic chain, and for its projection on the support plane (center of pressure), has been implemented to achieve easy posture control. Thus, the inverse kinemat- ics can be solved taking into account the static balance of the digital human. Moreover, the proposed algorithm allows sim- ulating quite complex tasks, which involve the motion of the whole body, by means of only few task-related control points arbitrary located on the whole kinematic chain. The resulting movements are quite natural even for complex tasks as can be seen on the simulation experiments reported which show the effectiveness of the proposed approach. Finally, the joint torques can be computed thanks to the kinetostatics duality: the results are in accordance with biomechanical analyses. Keywords Digital humans · Human-like motion generation · Multiple-point kinematic control · Posture control · Kinetostatic duality · Low-back biomechanical analysis G. Di Gironimo (B ) · A. Tarallo IDEA Lab, DiME, University of Naples Federico II, Piazzale Tecchio 80, 80125 Naples, Italy e-mail: giuseppe.digironimo@unina.it L. Pelliccia · B. Siciliano PRISMA Lab, DIS, University of Naples Federico II, Via Claudio 21, 80125 Naples, Italy e-mail: bruno.siciliano@unina.it 1 Introduction Over the years, the so-called ergonomics and usability have became key product quality parameters, while an increasing attention has been devoted to the so-called human-centred design approach, even from the early stages of the design process [1–4]. In particular, a great deal of research has been done on modelling “digital humans” that could faithfully reproduce, in a digital environment, the movements and the behaviour of the human beings [5, 6]. This has given birth to the so-called digital human modelling (DHM) which led to the develop- ment of many software tools [7, 8]. These tools are used for instance to study human–product and human–process interaction as well as to perform ergo- nomic and biomechanical analyses and even manual pro- cesses simulations. The digital humans essentially are kinematic chains con- sisting of several segments (links) connected by joints. The lengths of the segments, as well as their mass distribution, are derived from anthropometric databases, (e.g. [9, 10]) which can be queried with respect to different percentiles in the population. With reference to the “animation” of the digital human, commercially available software tools generally follow a “key-frame” approach that can be very time consuming, because of the difficulty to achieve, in this way, a faithful “behaviour” for the virtual humanoid. In fact, although they do implement effective inverse kine- matics (IK) algorithms, they can generally manage only a limited number of segments at time, that move independently of each other, without any care for balancing or other biome- chanics issues. As a result, the accuracy of those simulations is strongly dependent on the biomechanics skills of the oper- ator and his experience. 123