Perspectives in Autonomous Systems Research

Today, many devices (e.g., cars and other vehicles) we operate for various tasks (e.g., to go from place A to place B) are changing: in the past, they were characterized by a body and control actuators that allowed us to perform these tasks. These days, they are not simply passive recipients of our instructions; rather, they have become proactive tools performing the assigned task (or parts of it) by making autonomous decisions based on signals from their sensors. Therefore, more of these devices can be classified as autonomous systems. 

 

The level of autonomous decision making can be device dependent but, in every case, involves efficient multimodal signal processing capabilities. Such devices become our agents and have to relate multisensorial data representations to their actions and to the action context. Such a context includes the target of the agent’s task, i.e., the action motivation and the user, in cases in which the task is only partially automated.