Reference Number: INTAS-BELARUS-97-2028
|| Intelligent Neural System for Autonomous Control of a Mobile Robot
The objectives of the proposed project are:
- To develop an intelligent neural system for the movement control of a mobile robot.
- To create an "artificial" brain for control of both movement and
orientation of the robot in space. The general model of interaction
of the offered system with environment is shown in the following fig.
- To consider sensor inputs, providing the source information for
such a control system: diagram of environmental conditions from
range sensor devices; dead reckoning based on wheel encoders (coordinates of initial
and final point of robot movement is given). The sensors to be
employed will be low'cost ultrasonic and infrared sensors as well as cameras.
- To combine the information of multiple inexpensive sensors by data fusion
and adaptive control schemes to derive a fault tolerant, robust performance
of the robot. In particular when inexpensive sensors will be combined by
intelligent algorithms to derive efficient reactions, there can be expected
an interesting economic potential.
- To generate with the neural network the direction and speed of movement,
on the basis of the information from sensor devices. If necessary it should
be able to correct the internal map of the environment during the movement.
- To investigate the various neural nets for the different functionalities,
which are to be combined in an intelligent system.
- To solve the following problems with the neural system: navigation of the
robot for movements in only partially known environments; formation of a
region map during driving in the unknown area; planning of an optimum route
for the movement; supervised training; unsupervised training for self-organisation
during the movement of the robot; capability for real time actions.
Designed by Andrey Dunets
The Laboratory of Artificial Neural Networks
Brest State Technical University, Belarus