DRAMA, a Connectionist Architecture for Control and Learning in Autonomous Robots

Abstract

Adaptation to their environment is a fundamental capability for living agents, from which autonomous robots could also benefit. This work proposes a connectionist architecture, DRAMA, for dynamic control and learning of autonomous robots. DRAMA stands for dynamical recurrent associative memory architecture. It is a time-delay recurrent neural network, using Hebbian update rules. It allows learning of spatio-temporal regularities and time series in discrete sequences of inputs, in the face of an important amount of noise. The first part of this paper gives the mathematical description of the architecture and analyses theoretically and through numerical simulations its performance. The second part of this paper reports on the implementation of DRAMA in simulated and physical robotic experiments. Training and rehearsal of the DRAMA architecture is computationally fast and inexpensive, which makes the model particularly suitable for controlling computationally-challenged robots. In the experiments, we use a basic hardware system with very limited computational capability and show that our robot can carry out real time computation and on-line learning of relatively complex cognitive tasks. In these experiments, two autonomous robots wander randomly in a fixed environment, collecting information about its elements. By mutually associating information of their sensors and actuators, they learn about physical regularities underlying their experience of varying stimuli. The agents learn also from their mutual interactions. We use a teacher-learner scenario, based on mutual following of the two agents, to enable transmission of a vocabulary from one robot to the other.

Design and Application of an Online English Self-learning Platform

“…When building this kind of platform, we need to consider the following requirements [12][13][14][15][16][17]:…”

Play, Dreams and Imitation in Robota

“…I exploited this idea in experiments in which the robot's ability to imitate another agent, robot or human, was used to teach it a synthetic proto-language (Billard 2000a;Billard & Dautenhahn 1999;Billard & Hayes 1999). There, movement imitation is an implicit means of teaching.…”

“…These are used by the PC interface program to learn a language, that is, to extract words of the sentences and attach them a meaning in terms of the robot's perceptions (sensor and actuator states of the robot are passed to the PC through the serial link), see (Billard 2000a;Billard & Hayes 1999) for an explanation of the principle. The learning algorithm is provided by an artificial neural network, called DRAMA (Billard & Hayes 1999), which has general properties for learning time series. Once the language is learned, the robot can talk with the user through the PC loudspeaker.…”

Development of a miniature pair of eyes with camera for the humanoid robot Robota

“…Motor Board and Sensor Board are controlled by two PIC16FA microcontrollers. Schematic of the electronic boards and of the mechanics are available in the Robota UserGuide at http : //robota.epf l.ch Robota's behaviors allow multi-modal human-robot interactions, including the ability to imitate gesture, to dialog via speech processing/synthesis, and to learn simple composition of motions and of words, see [6,10].…”

“…Los Angeles, November [10][11][12]2004 robots have at most 1 "eye", i.e. one camera mounted directly in the head, the head having at most 2 DOFs.…”