Community Research and Development Information Service - CORDIS

Virtual imitation of human movement

During movement, large-scale muscular-skeletal systems are capable of being controlled, but the ways by which this is realised, have so far been difficult to simulate. Challenged by this, a Bulgarian research institute proposes an integrated approach for control design in computer simulation of both constrained and unconstrained movements. This approach may be of valuable importance for many fields, such as biomechanical engineering, human animation, virtual reality and robotics.
Virtual imitation of human movement
Human movement, the end-result of a synergistic activity between muscles and joints, is highly dependent on extremely complicated multi-body dynamics. Walking, running, jumping, grasping, or throwing are simple movements interrelated by a continuous sequence of complex changes of place and body posture. In all cases, muscular-skeletal systems are kept entirely controlled through potentially inexplicit mechanisms that need to be further explored in order to realistically simulate human movement.

Motivated by this, this Bulgarian research group developed a comprehensive strategy for the control design in computer simulation of human voluntary and involuntary movements. Involuntary movements, unconstrained in terms of space-time or force-time, the approach involves a control learning method that exploits repeated dynamics simulation and motion evaluation. On the other hand, in constrained movements, such as for posture tracking, suitably robust and optimal feedback controllers may be used for simulating human motions.

The extremely efficient methods for simulating multi-body biomechanics systems have been verified using two different case studies that both involves models of both constrained and unconstrained movements. The varying structure and parameters of the control system used allowed optimisation of the performance of simulated movement, such as execution time and control energy. In addition, the suitability of developed control-learning algorithm has been tested and assessed using computer simulation of a full dynamic model.

Employing realistic dynamics models in motion simulation of articulated structures may facilitate the development of new or improvement of existing control techniques used in prosthetic devices. The adoption of the offered approach may contribute significantly in the development of movement strategies and tendon, muscle and joint forces in humans performing specific manual tasks. Other useful applications may be found in the area of ergonomic optimisation of tools, workplaces and man/machine interfaces.
Record Number: 80842 / Last updated on: 2005-09-18
Domain: IT, Telecommunications