Autonomy is the ability to make independent choices or decisions and to determine the own actions (self-determination). It is also closely related to freedom (from all external constraints) and independence (from other systems, objects or entities). Autonomy is a property of systems with Self-Star Properties. The behavior of autonomous systems does not depend completely on external guidance and control.
Etymology: from Greek (griech.: αυτονομία, (αὐτονομία) autonomía = self-ruling or self-governing). It is a combination of Greek autos (self) and nomos (law), and means therefore literally a system which rules itself.
Self-Determination and Context-Dependence
The context is very important for autonomous systems. Contrary to machine parts or traditional software objects, an autonomous system has usually no fixed role or unchangeable function, except the role it currently occupies. Although it is autonomous, it can be context-dependent, and change its role according to influences from the environment. This change can be deliberately or unconsciously. An ant in an ant colony for example can occupy the role transporter, explorer, soldier, etc. Context-dependence of autonomous (sub-)systems is especially important for self-organizing systems.
Reactive and Proactive
An autonomous system can not be purely reactive. At least parts of the behavior should be proactive. Proactive action is active, anticipatory and goal-directed action to reach a certain objective or to prevent a particular situation. Contrary to reactive behavior, which is a passive action determined by a fixed input-output (I-O) relation, proactive behavior is an active action determined by a output-input (O-I) relation. Reactive behavior is a reaction to past events, proactive behavior is an anticipatory action to influence future events.
Autonomy is a typical property of robots (autonomous vehicles, undersea robots, mars exploration rovers like Spirit and Opportunity, etc.) and agents. Autonomous robots are robots which can perform desired tasks in unstructured environments without continuous human guidance and humans "in the loop".