An Software Architecture Design For AI Service Robot

Since1950s the industrial robot has developed itself be widely used in manufacture and assembly line to help workers to accomplished a huge amount of repetitive and laborious jobs. However, the current employed industrial robots could only operate in the preset worksohp environment, do little help to the normal life of common people, who hold enthusiastic dreams about the intelligent robots. The industrial robots could not yet handle with the y office environment or home setting, neither could they cooperate with the humankind to accomplish a single job. On the other side, people’s appeal for daily use of service robots in ordinary social environment is blooming, because of the aging society and the wide spreaded using of electronic productions.There are mainly two obstacles to hurdle if we wish to introduce the service robot into the popular life. The first is the human-robot interaction, and the second is the capability of robots to adapt themselves to the complex and easy-changing environment. In according to solve these problems, researchers endeavor briefly in two fields, software and hardware, to enforce the robots to be more flexible and powerful. A robot based on an intelligent software system and well deviced sensing acuity could provide the permission to allow the non-expert users to interactive with the robot, as well as competence to fit itself in the home setting and office environment. Besides, an intelligent software system also endows the robots the ability of implementing complex structured tasks. The research in hardware field, particularly in the field of mechanical design and control, as well as transducer, helps the robots to move themselves in such a complex environment more skillfully.This paper is dedicated to furnish the robots a more intelligent software system. In this paper are firstly introduced the historical and cultural background of robots and artificial intelligence, to convey our design stance of the service robots. Thereafter we summarized four different AI robot design paradigms which were developed or yet being explored in the past forty years’ of AI robot history:hierarchical paradigm, reactive paradigm, hybrid paradigm, and learning paradigm. Our intelligent service robot software architecture, RobWen, are grounded on the hybrid paradigm and learning paradigm, involving five main function modules:planner, perceptor, actor, cognitive module and memory. RobWen performs the tasks regularly based on the decisions from the planner. With the help of sensor and environment analyzer RobWen could move itself in the office and campus environment setting.The software architecture of our robot is grounded on the distributed robot API, Player. Under the architecture of Player, the robot carries out its various tasks concurrently in the special threads. The messages between threads offer the cooperative mechanism of different modules to accomplish the complex tasks. In the last section we provide two examples of task implementing on the robot simulation platform, Stage, through which we proved that the methodology of building such a robot is feasible.