Abstract
The high speed of saccades means that they cannot be guided by visual feedback, so that any saccadic control system must know in advance the correct output signals to fixate a particular retinal position. To investigate neural-net architectures for learning this inverse-kinematics problem we simulated a 4 deg-of-freedom robot camera-head system, in which the head could pan and tilt and the cameras pan and verge. The main findings were: (1) Linear nets, multilayer perceptrons (MLPs) trained by backpropagation, and cerebellar model arithmetic computers (CMACs) all learnt rapidly to 5–10% accuracy when given perfect error feedback. (2) For additional accuracy (down to 2%) two-layer nets learnt much faster than a single MLP or CMAC: the best combination tried was to have a CMAC learn the errors of a trained linear net. (3) Imperfect error signals were provided by a crude controller whose output was simply proportional to retinal input in the relevant axis, thereby providing a mechanism for (a) controlling the camera-head system when the feedforward neural net controller was wrong or inoperative, and (b) converting sensory error signals into motor error signals as required in supervised learning. It proved possible to train neural-net controllers using these imperfect error signals over a range of learning rates and crude-controller gains. These results suggest that appropriate neural-net architectures can provide practical, accurate and robust adaptive control for saccadic movements. In addition, the arrangement of a crude controller teaching a sophisticated one may be similar to that used by the primate saccadic system, with brainstem circuitry teaching the cerebellum.
Similar content being viewed by others
References
Albus JS (1971) A theory of cerebellar function. Math Biosci 10:25–61
Albus JS (1981) Brains, behavior and robotics. BYTE books. Mc-Graw-Hill, Peterborough New Hampshire
Anastasio TJ, Robinson DA (1989) The distributed representation of vestibulo-oculomotor signals by brain-stem neurons. Biol Cybern 61:79–88
Aslin RN (1987) Motor aspects of visual development in infancy. In: Salapatek P, Cohen L (eds) Handbook of infant perception. Academic Press, New York, pp 43–113
Atkeson CG (1989) Learning arm kinematics and dynamics. Ann Rev Neurosci 12:157–183
Broomhead DS, Lowe D (1988) Multi-variable functional interpolation and adaptive networks. Compl Sys 2:269–303
Brown C (1990) Prediction and cooperation in gaze control. Biol Cybern 63:61–70
Carpenter RHS (1988) Movement of the eyes, 2nd ed. Pion, London
Carpenter GA, Grossberg S (1987) A massively parallel architecture for a self-organising neural pattern recognition machine. Comput Vision Graph Image Proc 37:54–115
Crick F (1989) The recent excitement about neural networks. Nature 337:129–132
Eisley RK (1988) A learning architecture for control based on back-propagation neural networks. IEEE Int Conf Neural Networks 2:587–594
Fu KS, Gonzalez RC, Lee CSG (1987) Robotics: control, sensing, vision, and intelligence. McGraw-Hill, New York
Funahashi K-I (1989) On the approximate realization of continuous mappings by neural networks. Neural Networks 2:183–192
Grossberg S (1987) Competitive learning: from interactive activation to adaptive resonance. Cog Sci 11:23–63
Grossberg S, Kuperstein M (1986) Neural dynamics of adaptive sensory-motor control: ballistic eye movements. In: Stelmach GE, Vroon PA (eds) Advances in psychology, vol 30. North-Holland, Amsterdam
Guez A, Ahmad Z (1988) Solution to the inverse kinematics problem in robotics by neural networks. IEEE Int Conf Neural Networks 2:617–624
Kawato M (1989) Neural network models for formation and control of multijoint arm trajectory. In: Ito M (ed) Neural programming. Taniguchi Symposia on Brain Sciences No 12. Japan Scientific Society Press/Karger, Basel, pp 189–201
Kawato M (1990) Feedback-error-learning neural network for supervised motor learning. In: Eckmiller R (ed) Advanced Neural Computers. Elsevier, Amsterdam, pp 365–372
Kawato M, Furukawa K, Suzuki R (1987) A hierarchical neural-network model for control and learning of voluntary movement. Biol Cybern 57:169–185
Keller EL (1989) The cerebellum. In: Wurtz RH, Goldberg ME (eds) The neurobiology of saccadic eye movements. Elsevier, North Holland, pp 391–411
Kuperstein M (1988a) An adaptive neural model for mapping invariant target position. Behav Neurosci 102:148–162
Kuperstein M (1988b) Neural model of adaptive hand-eye coordination for single postures. Science 239:1308–1311
Lapedes A, Farber R (1987) Nonlinear signal processing using neural networks: prediction and system modelling. Los Alamos National Laboratory, Preprint LA-UR-87-2662
Marr D (1969) A theory of cerebellar cortex. J Physiol 202:437–470
Miller WT III, Glanz FH, Kraft LG III (1987) Application of a general learning algorithm to the control of robotic manipulators. Int J Robot Res 6:84–97
Miyamoto H, Kawato M, Setoyama T, Suzuki R (1988) Feedbackerror-learning neural network for trajectory control of a robotic manipulator. Neural Networks 1:251–265
Moody J (1989) Fast learning in multi-resolution hierarchies. In: Touretzky DS (ed) Advances in neural information processing systems. Morgan Kaufman, San Mateo, CA, pp 29–39
Optican LM (1982) Saccadic dysmetria. In: Lennestrand G, Zee DS, Keller EC (eds) Functional basis of oculomotor motility. Pergamon Press, Oxford, pp 441–451
Parker D (1985) Learning logic. Technical Report TR-87, Center for Computational Research in Economics and Management Science, MIT, Cambridge, Mass
Pellionisz A (1984) Coordination: a vector-matrix description of transformations of overcomplete CNS coordinates and a tensorial solution using the Moore-Penrose generalized inverse. J Theor Biol 101:353–375
Ritter H, Schulten K (1988) Extending Kohonen's self-organizing mapping algorithm to learn ballistic movements. In: Eckmiller R, von der Marisburg C (ed) Neural Computers, NATO ASI Series, vol F41. Springer, Berlin Heidelberg New York pp 393–406
Rummelhart DE, Hinton GE, Williams RJ (1986) Learning internal representations by error propagation. In: Rummelhart DE, Mc-Clelland JL (eds) Parallel distributed processing: explorations in the microstructures of cognition. MIT Press, Cambridge Mass, pp 318–362
Snyder WE (1985) Industrial robots: computer interfacing and control. Prentice-Hall, Englewood Cliffs NJ
Thacker NA, Mayhew JEW (1990) Designing a layered network for context sensitive pattern classification. Neural Networks 3:291–299
Tweed D, Vilis T (1987) Implications of rotational kinematics for the oculomotor system in three dimensions. J Neurophys 58:832–849
Waxman AM, Wong W-L, Goldenberg R, Bayle S, Baloch A (1988) Robotic eye-head-neck motions and visual-navigation reflex learning using adpative linear neurons. Neural Networks 1 [Suppl 1]:365
Werbos PJ (1974) Beyond regression: new tools for prediction and analysis in the behavioural sciences. PhD thesis, Harvard University, Cambridge, Mass
Widrow B, Hoff ME (1960) Adaptive switching circuits. In: 1960 IRE WESCON Convention Record. IRE, New York, pp 96–104
Widrow B, Stearns SD (1985) Adaptive signal processing. Prentice-Hall Engelwood Cliffs NJ
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Dean, P., Mayhew, J.E.W., Thacker, N. et al. Saccade control in a simulated robot camera-head system: neural net architectures for efficient learning of inverse kinematics. Biol. Cybern. 66, 27–36 (1991). https://doi.org/10.1007/BF00196450
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF00196450