- Full title Using multisensor fusion and neuronal network techniques for robot control
- Funding entity OTAN
- Researchers Fernando López Peña, Francisco Javier Bellas Bouza, Richard J Duro Fernández, Adolfo Lamas Rodríguez, José Antonio Becerra Permuy, Juan Luis Crespo Mariño
- Start date 01/06/2002
- End date 01/06/2003
Collaborating entities
- Ternopil Academy of National Economy
- Universidade da Coruña
Description
In this project, we have developed and implemented some of the major concepts for sensor fusion. To this end, neural networks for controlling mobile robots have been used. It is assumed that the robot’s movement takes place in undefined areas, which are probably dangerous for humans. Another assumption is that the robot´s main objective is to achieve a certain point in an area that contains obstacles.
The study is based on neural computing techniques that provide appropriate control strategies for mobile robots with uncertain sensing and use their capacity for self-organisation and self-adjustment.
We have achieved the following results:
- An innovative sensor fusion technique based on neural networks.
- Algorithms and sensors for a sensor fusion subsystem.
- Hardware implementation of the sensor fusion technique in a mobile robot.
- Procedures for testing and verifying the sensor fusion system.