To circumvent this shortcoming, we propose a behavioristic robotic system that not only guides a social robot to explore a home environment with only modest prior knowledge but also builds a local map and registers its location within that map. Additionally, the conventional behavioristic design often acts blindly and does not explicitly include a learning module that is essential for the robot (agent) to function in a purposeful manner.
It could be argued that this strategy is more challenging than a purely deliberative design that depends on prior planning and an accurate map. In this paper, we report a methodology within the general behavioristic architecture (Sense-Act system). Whereas the former has been extensively studied with successful implementations the interest in the latter, after a dormant period, has been revived in the last decade through advancements of deep learning and problem-solving algorithms inspired by the natural world. This problem has been studied within the context of probabilistic and behavioristic paradigms, respectively. Hence, designing a robotic navigation system with limited sensors that is capable of exploring an indoor environment while the robot concurrently updates its pose and generates/updates its map is an active research area in robotics. It is, thus, sensible to suggest that such skills can hardly be isolated from the crucial ability to explore, navigate, or perceive information in the way humans do. Indeed, a social robot cannot realistically perform its dedicated tasks, if it is immobile.
Implicit but an indispensable feature of a social robot is its ability to seamlessly move around in the environment for which it is expected to function. Depending on its ultimate application, a social robot can be designed as a pet-like, e.g., AIBO, or a humanoid, e.g., Nao, or a wheeled robot, e.g., Pepper, or unmovable robot, e.g., Kasper. Social robots are designed in all sizes and shapes for a variety of applications but most importantly, they are designed to be acceptable to humans. Other key features of such robots are their ability to recognize people, objects, communicate through voice, and respond to various human emotions. Social robots are designed for a variety of tasks in a collaborative or service setting and could be deployed in homes (to do household chores, act as a companion to children and seniors, or serve as a butler, etc.), hospitals (as a nurse, administrative assistant, etc.), schools (as a teacher), libraries (as a librarian), museums (guides, etc.), to name a few. As such, avatars and virtual agents are generally excluded. Embodiment is an essential characteristic of this class of robots. Social robots are referred to as a special family of autonomous and intelligent robots that are predominantly designed to interact and communicate with humans or other robots (agents) within a collaborative environment.
0 kommentar(er)
