MURPHY: A Robot that Learns by Doing
نویسنده
چکیده
MURPHY consists of a camera looking at a robot arm, with a connectionist network architecture situated in between. By moving its arm through a small, representative sample of the 1 billion possible joint configurations, MURPHY learns the relationships, backwards and forwards, between the positions of its joints and the state of its visual field. MURPHY can use its internal model in the forward direction to "envision" sequences of actions for planning purposes, such as in grabbing a visually presented object, or in the reverse direction to "imitate", with its arm, autonomous activity in its visual field. Furthermore, by taking explicit advantage of continuity in the mappings between visual space and joint space, MURPHY is able to learn non-linear mappings with only a single layer of modifiable weights.
منابع مشابه
Obstacle Avoidance by Means of an Operant Conditioning Model
This paper describes the Clpplication of a model of operant conditioning to the prohlcrn of obstacle avoidance \Vith a wheeled mobile robot. The main characteristic of the 8pplicd model is that the robot learns to avoid obstacles through a learning-by-doing cycle without external supervision. A series of ultrasonic sensors aetas Conditioned Stimuli (CS), while collisions act as an Unconditioned...
متن کاملToward Teaching a Robot "Infant" using Emotive Communication Acts
This paper presents ongoing work towards building an autonomous robot that learns in a social context. The mode of social interaction is that of a caretaker-infant pair where a human acts as the caretaker for the robot. By placing our robot, Kismet, in an environment with a human caretaker who actively assists and guides Kismet's learning, this work explores robot learning in a similar environm...
متن کاملVision Based Robot Navigation
In this thesis we propose a vision-based robot navigation system that constructs a high level topological representation of the world. A robot using this system learns to recognize rooms and spaces by building a hidden Markov model of the environment. Motion planning is performed by doing bidirectional heuristic search with a discrete set of actions that account for the robot's nonholonomic con...
متن کاملNeural Controller for a Mobile Robot in a Nonstationary Environment
Recently it has been introduced a neural controller for a mobile robot that learns both forward and inverse odometry of a diierential-drive robot through an unsupervised learning-by-doing cycle. This article introduces an obstacle avoidance module that is integrated into the neural controller. This module makes use of sensory information to determine at each instant a desired angle and distance...
متن کاملFurther Explorations in Visually-Guided Reaching: Making MURPHY Smarter
MURPHY is a vision-based kinematic controller and path planner based on a connectionist architecture, and implemented with a video camera and Rhino XR-series robot arm. Imitative of the layout of sensory and motor maps in cerebral cortex, MURPHY'S internal representations consist of four coarse-coded populations of simple units representing both static and dynamic aspects of the sensory-motor e...
متن کامل