Teaching a Robot with Human Natural Movements
A human inspired sensorimotor approach to robotic systems could solve the problem of designing adaptive and learning robot for widely versatile tasks. We present a neural network controller for driving a 3 degree of freedom mechanical robotic finger to a desired position in space. The controller has been taught with human movements of the corresponding finger. At the end of the training process, it was able to closely imitate the physiological control and the motion planning strategy of the human beings. Generalization properties are shown after the training process (that is the capacity to move the device in different directions and places never seen during the teaching phase). The approach seems promising for controlling artificial prosthetic and robotic upper limbs in general.
KeywordsKinematic Model Humanoid Robot Multi Layer Perceptron Real Movement Natural Movement
Unable to display preview. Download preview PDF.
- Flash, T., and Hogan, N. (1985). The coordination of arm movements: an experimentally confirmed mathematical model. Journal of Neuroscience 5:1688–1703.Google Scholar
- Secco, E.L., and Magenes, G. (2002). A feedforward neural network controlling the movement of a 3 degree of freedom finger, IEEE SMC Transaction Part A 32(3):437–445.Google Scholar
- Secco, E.L., Visioli, A., and Magenes, G. (2002). Minimum jerk motion planning for a prosthetic finger, Journal of Robotic Systems (submitted).Google Scholar
- Vicini, O. (2003). Controllo neurale di un dito robotico allenato tramite movimenti di un dito umano, Thesis, Tutor Magenes, G., Dip. Informatica e Sistemistica, University of Pavia, Pavia — ItalyGoogle Scholar
- Caiti, A., Canepa, G., De Rossi, D., Germagnoli, F., Magenes, G., Parisini, T. (1995). Towards the realization of an artificial tactile system: fine-form discrimination by a tensorial tactile sensor array and neural inversion algorithms. IEEE SMC Transaction 25(6):933–946.Google Scholar
- Magenes, G., Ramat, S., Secco, E.L. (2002). Life-like sensorimotor control: from biological systems to artifacts. Current Psychology of Cognition 21(4–5): 565–596.Google Scholar