Recent years have witnessed the consolidation of a new approach to research into the higher functions performed by the nervous systems. The biological approach, based upon information deriving from fMRI studies, single unit recordings, lesion studies on primates and other branches of scientific research, was recently integrated and complemented with the biomedical engineering and robotics approach, providing an invaluable test bed for new hypotheses and theories. A radically new approach to the problem of designing and producing a general purpose humanoid robot able to interact intelligently with the environment has been proposed in the last decade. Inspired by biological systems, the behavior-based approach avoids higher-level representation and relies upon a tight connection between sensory and motor systems. From these grounds evolved the assumption that, to have human-like capabilities, robots should have human-like senses and bodies. The hypothesis is supported by the observation that, on one hand, the solution found by nature is the outcome of millions of years of evolution and that, on the other hand, the intended artifacts can develop a human-like intelligence only through human-like interaction with the environment. In this work we present two models of human sensorimotor subsystems which can be considered as examples of building blocks for the realization of a robot capable of intelligent behavior. The first is a model of eye-head coordination, which, through learning, is able to adapt the eye-head coordination behavior to different conditions requiring the visual exploration of the surroundings. The second example concerns the haptic sensitivity, which allows humans to explore the environment through the sense of touch. We developed an artificial tactile system combining passive touch with active finger motion. Tactile sensors and finger movement control should allow an artifact to perform haptic exploration of objects and object recognition through stereognosis
Life-like sensorimotor control: from biological systems to artifacts
MAGENES, GIOVANNI;RAMAT, STEFANO;SECCO, EMANUELE LINDO
2002-01-01
Abstract
Recent years have witnessed the consolidation of a new approach to research into the higher functions performed by the nervous systems. The biological approach, based upon information deriving from fMRI studies, single unit recordings, lesion studies on primates and other branches of scientific research, was recently integrated and complemented with the biomedical engineering and robotics approach, providing an invaluable test bed for new hypotheses and theories. A radically new approach to the problem of designing and producing a general purpose humanoid robot able to interact intelligently with the environment has been proposed in the last decade. Inspired by biological systems, the behavior-based approach avoids higher-level representation and relies upon a tight connection between sensory and motor systems. From these grounds evolved the assumption that, to have human-like capabilities, robots should have human-like senses and bodies. The hypothesis is supported by the observation that, on one hand, the solution found by nature is the outcome of millions of years of evolution and that, on the other hand, the intended artifacts can develop a human-like intelligence only through human-like interaction with the environment. In this work we present two models of human sensorimotor subsystems which can be considered as examples of building blocks for the realization of a robot capable of intelligent behavior. The first is a model of eye-head coordination, which, through learning, is able to adapt the eye-head coordination behavior to different conditions requiring the visual exploration of the surroundings. The second example concerns the haptic sensitivity, which allows humans to explore the environment through the sense of touch. We developed an artificial tactile system combining passive touch with active finger motion. Tactile sensors and finger movement control should allow an artifact to perform haptic exploration of objects and object recognition through stereognosisI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
