Collaborative Autonomous Learning Systems
In many applications there may be more than one systems that can act autonomously. There are numerous examples of such situations. For example, students and teachers in a class room, drivers and pedestrians on the roads, and so on. These are natural systems. Similarly, one can imagine artificial autonomous systems (robots, agents). For example, a team of uninhabited vehicles (aerial, ground, water or underwater), a sensor network in which each node may be an intelligent sensor, enabled with processing and communication capability, ensemble of classifiers that may just be a software or including some hardware realisation (e.g. image-based), and so on. In such scenarios there may be, generally, the following two modes in which these autonomous entities work, as follows:
In what follows the collaborative scenario will be briefly described as applied to ALS in the form of clustering, classifiers, controllers, predictors, estimators, filters or intelligent sensors.
10.1 Distributed Intelligence Scenarios
In a collaborative scenario, each ALS acts on its own pursuing its own objectives and goals, but they can collaborate to achieve a common, shared goal. For example, in the so-called self-localisation and mapping (SLAM) problem (Choset and Nagatani, 2001) each mobile robot from a team can act on its own to localise ...