Intelligent and distributed control scenario for the fast and easy deployment of robots in divers...

Intelligent and distributed control scenario for the fast and easy deployment of robots in diverse environments

A fast and easy deployment of robots in new areas is necessary to get robots operating outside research centres and beyond the continuous supervision of roboticists. Robots must be capable of being installed and put into operation within very short periods of time. This is our main aim in this project and, to achieve it, we must face important challenges:

Complete avoidance of pre-definition, learning from experience and human observation: Like humans, robots should be able to learn from their own experiences when emulating people or exploring an environment. The mistakes and successes the robot makes should influence its future behaviour rather than relying only on predefined rules, models or hardwire controllers. This will result in robots that are able to adapt and change according to the environment. No pre-defined knowledge will be used to control the robot. Most of the robot’s competences will be learned through direct physical interaction with the environment and human observation.

Scene recognition capabilities: Today’s robots are unable to understand their environments, they are not aware if they are moving in a room that is similar to another one where they have been moving previously. Robots are not able to distinguish whether they are moving for instance in a kitchen or living room. On the other hand, the automatic detection of representative situations that require the presence of robots – a room without people that can be tidied up, people sitting in a sofa or children playing, people that have just entered home and can be welcomed, among others - represents an important qualitative leap forward as robots would stop from being passive and transform into robots with “initiative”.

Exploring strategies to identify and locate people and robots in indoor environments. Allowing people and robots to discover and communicate their positions is important if they want to interact among themselves.

These challenges will be addressed as part of the design of a "distributed intelligent control scenario". Considering future intellectual property exploitation, our control scenario must:

Allow a fast deployment of small and large robots in different environments.
Be friendly with the user.
Be scalable (it has to allow the introduction of new robots).

A fast and easy deployment of robots in new areas is necessary to get robots operating outside research centres and beyond the continuous supervision of roboticists. Robots must be capable of being installed and put into operation within very short periods of time. This is our main aim in this project and, to achieve it, we must face important challenges: <ol><li> Complete avoidance of pre-definition, learning from experience and human observation: Like humans, robots should be able to learn from their own experiences when emulating people or exploring an environment. The mistakes and successes the robot makes should influence its future behaviour rather than relying only on predefined rules, models or hardwire controllers. This will result in robots that are able to adapt and change according to the environment. No pre-defined knowledge will be used to control the robot. Most of the robot’s competences will be learned through direct physical interaction with the environment and human observation. </li> <li>Scene recognition capabilities: Today’s robots are unable to understand their environments, they are not aware if they are moving in a room that is similar to another one where they have been moving previously. Robots are not able to distinguish whether they are moving for instance in a kitchen or living room. On the other hand, the automatic detection of representative situations that require the presence of robots – a room without people that can be tidied up, people sitting in a sofa or children playing, people that have just entered home and can be welcomed, among others - represents an important qualitative leap forward as robots would stop from being passive and transform into robots with “initiative”.</li> <li>Exploring strategies to identify and locate people and robots in indoor environments. Allowing people and robots to discover and communicate their positions is important if they want to interact among themselves.</li></ol> These challenges will be addressed as part of the design of a "distributed intelligent control scenario". Considering future intellectual property exploitation, our control scenario must:<ul><li>Allow a fast deployment of small and large robots in different environments.</li><li>Be friendly with the user.</li><li>Be scalable (it has to allow the introduction of new robots).</li></ul> - Roberto Iglesias Rodríguez - Eva Cernadas García, Xosé Manuel Pardo López, Manuel Fernández Delgado