A multi-agent method and system for moving a virtual dummy in a virtual environment, the dummy being defined by an overall position and by a plurality of degrees of freedom for joints. The method includes a contribution from an attraction agent acting on the plurality of degrees of freedom of the joints of the dummy to move the dummy towards the target; and a contribution from an avoider agent acting on the overall position of the dummy as a function of parameters defining the environment in order to avoid collisions between the dummy and said environment. The method further includes a contribution from an ergonomy agent acting on the plurality of degrees of freedom of the joints of the dummy in order to correct the posture of the dummy automatically while it is moving towards the target.