An Implementation Of The Role Based Robot Perception Framework

With the rapid development of robot technology,different platforms and different types of robots emerge in an endless stream.At present,robots have been widely used in industry,and gradually extended to home,service,military and other fields.As an important component of robot's perception of external information,sensors play an important role in the task of robot perception.Different robot equipped with different sensors,provides different application development interface,the same perception task needs to adapt to different robots,the interface is not universal which greatly increase the workload of developers.It is meaningful to manage sensors of robots effectively,and provide a uniform interface for the development of robot perception tasks.It is significant to realize interconnection and operation among different robots for the development of robot perception tasks.A robot usually carry out perception tasks in complex context.In order to improve the environmental adaptability when performing the perception tasks,many researchers and practitioners developed many context-aware applications.Highly coupled context recognition module and perception service module is not conducive to the rapid construction of task and system implementation.It is one of the challenges for the researchers and practitioners in the field of Robot Perception to decouple the context aware and perceptual services to support the rapid construction of robot context aware applications.This paper focuses on pervasive sensor management of robot and the support of the rapid construction of context aware applications in open environments based on the open source robot operating system ROS.We proposed a role based robot sensor management model and a context driven robot perception computing framework,which together constitute the role based perceptual framework.The work of this paper includes:Firstly,on the base of the deep research in the robotic middleware and the sensor management module of the most popular mobile phone operating system Android,we concise the common needs in sensor management of robot.We classified sensor and propose the concept of sensor role,and design the data acquisition plug-in components of sensor role based on ROS which provide a unified interface for robot sensor data acquisition of different types of robot.Secondly,on the base of the deep research of context aware computing framework in the pervasive computing filed,we concise the common needs in context aware perception computing framework in robot filed.We focus on the context recognition and the intelligent actuator construction,design and implement a context driven perception computing framework,which decouple context recognition and perception services.The framework support the rapid design and prototype implementation of perception tasks.Thirdly,the underlying sensor management module and the upper context aware computing framework constitute the role based perception framework.We implemented the framework on the ROS,and construct the sensor management model for Turtlebot mobile robot.Then we constructed and implemented a typical context aware application quickly through the framework.The validity of this work was verified.