» Home
ICT, Electronics and Optics Artificial Intelligence
Artificial Intelligence
Our expertise. iMEX.A expertise covers mainly the following areas:
|
The remote monitoring and diagnostics are two basic features for the development of automated systems and plants, they consist of the system capability to recognise, in an autonomous way, the possible presence of failure symptoms, operative malfunctions or deviations from expected behaviour and detect the origin thanks to an appropriate logic implemented in the control software. If a system is also able to select and implement, in an autonomous way, also the reinstatement of the correct behaviour, it is defined FDIR (Failure Detection, Identification or Isolation & Recovery) system.
Different techniques derived from the artificial intelligence can be applied in this field. Our expertise is focused mainly on model-based monitoring techniques, model generation by means of inductive learning, fuzzy logic and neural networks.
Knowledge management techniques and data mining are also relevant issues since their employment concerns any application where a large amount of data is available and only the necessary information must be extracted. Data mining methods that can be applied are multiple (inductive learning, clustering, probabilistic networks and Bayes network, possibility networks, etc.), with emphasis to those methods (e.g., fuzzy logic or possibility logic) which allow managing the uncertainties and/or imprecision and/or the incompleteness of the initial data in a reliable and robust way thanks to an appropriate formalism.
The industrial automation and the on-board autonomy in the area of robotics and aerospace, represent the typical scenario to develop supervision systems which are able to substitute the human operator or to assist him providing information on the status of the system and support a decision. The on-board supervisors are therefore embedded systems with the control logic and the transducers setup which assures the correct implementation of the operations needed to fulfil the objective, whether the system behaves in the nominal way or in case of detected failures.
The autonomous agents are the most evident example of the capability of the artificial intelligence since they represent intelligent operators which are able to fulfil a given mission in substitution of the human operator. They are not simple action performers, they are agents which are able to take decisions and choose and manage their own actions to reach specific objectives. The architecture of autonomous agents counts on on-board supervisor, a failure monitoring and management system, a planning and scheduling system and other systems required by the objectives.
Applications. The different areas of the artificial intelligence offer tools and techniques which are used in multiple applications with the objective to increase the automation or the autonomy of systems of different nature, to compensate the limits in the management of the operations by human operators and to supply an advanced and intelligent support to solve complex problems.
Some examples of application areas of artificial intelligence techniques are: mechanics, aerospace, medicine, food, entertainment, etc.
Our experience. iMEX.A developed an algorithm for the supervision and autonomous management of the docking phase between lunar rovers on ground, exploiting the formalism of the artificial intelligence. The study was developed in a two-year contract as consultants for Thales Alenia Space Italia within the STEPS project which deals with the development of new space technologies. The STEPS project is financed by Regione Piemonte.
The designed system allows control of every phase of the docking in order to verify that all the nominal working conditions are satisfied. If one or more of these conditions should not be verified due to adverse external causes, the algorithm was designed to react autonomously to fulfil the mission in an alternative way or to abort it safely without the need of human action.
 |
|