Adaptation to their environment is a fundamental capability for living agen
ts, from which autonomous robots could also benefit. This work proposes a c
onnectionist architecture, DRAMA, for dynamic control and learning of auton
omous robots. DRAMA stands for dynamical recurrent associative memory archi
tecture. It is a time-delay recurrent neural network, using Hebbian update
rules. It allows learning of spatio-temporal regularities and time series i
n 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 ar
chitecture and analyses theoretically and through numerical simulations its
performance. The second part of this paper reports on the implementation o
f DRAMA in simulated and physical robotic experiments. Training and rehears
al of the DRAMA architecture is computationally fast and inexpensive, which
makes the model particularly suitable for controlling 'computationally-cha
llenged' robots. In the experiments, ave use a basic hardware system with v
ery 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 fix
ed environment, collecting information about its elements. By mutually asso
ciating information of their sensors and actuators, they learn about physic
al regularities underlying their experience of varying stimuli. The agents
learn also from their mutual interactions. We use a teacher-learner scenari
o, based on mutual following of the two agents, to enable transmission of a
vocabulary from one robot to the other.