DOC PREVIEW
UNCC ECGR 6185 - RTOS-based Software Architecture for Multisensor Fusion System

This preview shows page 1-2-3 out of 8 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

RTOS-based Software Architecture for Multisensor Fusion System Zhang Kejun* and Su Jianbo * Department of Automation & Research Center of Intelligent Robotics Shanghai Jiaotong UniversityShanghai, China. e-mail:[email protected] this paper, software architecture for multisensor fusionsystem (MFS) in dynamic environment is described. Thearchitecture has the characteristics of robustness,dynamically reconfigurable framework, “plug-in-play” forsensor, and predictability of sensor action. According torequirements of MFS, the real-time operating system (RTOS) is applied in this software architecture. The resultis an RTOS-based software architecture, which consists of five kinds of tasks: Sensor Sampling Task (SST), SensorData Processing Task (SDPT), Fusion Task (FT),Management Task (MT) and Man/Machine Interface Task(MMIT). A calligraphic robot has been designed anddeveloped in our lab based on the software architectureproposed. The application of the architecture makes thesensor management of the calligraphic robot system easy. 1 Introduction Multisensor fusion system (MFS) is a complex informationprocessing system. Several sensors that have differentfunctions or spatial locations are integrated to get betterinformation of environment. MFS has been applied in manyfields such as military, satellite, aircraft navigation, robotsand industrial assembly [19]. Currently, the research ofMFS in dynamic environment has become an interestedarea [25]. In dynamic environment, a significant feature ofMFS is that the sensor sampling, data processing and fusionmust be finished in limited time. That is to say that real-time characteristic is an important factor for the availabilityof whole system. Normally, more sensors and complexhardware are applied to meet the requirement of real-time.Accordingly, with the workload of hardware managementincreasing dramatically, a good software structure is necessary to satisfy the real-time requirement of the wholesystem.In fact, the software architecture has been of great interestin the field of MFS integration. The concept of “logicalsensor” (LS) has been widely studied in MFS, which is aspecification for the abstract definition of a sensor that can be used to provide a uniform framework for multisensorsystem integration [1, 2, 3]. In [4], an object-orientedprogramming method is proposed to develop a uniformframework for implementing multisensor systems, in whicheach sensor is represented as an object and objectscommunicate by passing messages. Literature [5] describedthe software architecture of science/autonomy system (SAS)in which each sensor was represented by a sensor object.Jun Wang [6] proposed an open, module and hierarchicalsoftware architecture for MFS, where component objectmodel (COM) was used to build the whole softwarearchitecture. These methods facilitate system constitution,software packing and software re-usage. Unfortunately,they only focus on the efficiency of software design ratherthan the time constraint of system operating. MFS workingunder dynamic environment are required to be a real-timesystem. Thus time constraint is an additional importantaspect, which determines performance of the system [7]. Accordingly, it is necessary to introduce the real-timeoperating system (RTOS) to develop the MFS under thedynamic environment. An RTOS can be divided intoseveral parts, including the real-time kernel, the file system,and the programming environment. Among them, the real-time kernel is the key to endow a MFS with real-time feature. It provides local task management, scheduling,timing primitives, memory management, localcommunication, interrupt handling, error handling, and aninterface to hardware devices. The architecture of a practical RTOS is shown in Fig. 1. As shown in it, severalapplication tasks are running concurrently on the RTOS.Operating SystemSystem HardwareTasksFigure 1: The architecture of RTOS in application There are many kinds of RTOS. Every kind of RTOSs hasits own characteristics. The most famous one is Vxworks,which has a successful application in the Mars Pathfinder. This RTOS implements a time slice of severalmicroseconds [8]. QNX is another frequently used RTOS.In [9], the analysis and application of QNX were given.RT_Linux is another famous RTOS. Many applications ofrobot used RT_Linux [10]. All these OS are hard real-timeoperating system. Besides, Windows NT is also a kind ofRTOS, which is a soft real-time operating system, and haswidely been accepted as in developing applications [11]. This paper is to explore RTOS-based MFS softwarearchitecture, considering the requirements of MFS and the characteristics of RTOS, e.g. Windows NT. The proposedarchitecture is implemented in developing a calligraphicrobot, which shows its feasibility and validity.Section 2 discusses the requirements of the general MFS indetail, especially under dynamic environment. Section 3proposes a RTOS based software architecture for MFS,which is independent of the specific applications. Section 4shows a case study where a robot is used to write Chinesecharacters on the software architecture proposed, followedby the conclusions and future research. 2 Multi-sensor Fusion System 2.1 General PatternThe general pattern of MFS proposed in [12] has beenwidely accepted. It is illustrated in Fig. 2. The followingdiscussions are also based on this general pattern of theMFS.Sensorselect ionWorldmodelDat a trans-format ionSYSTEMEnvironment123nFusionFus i o nFusionInteractive system informationSensorsLevel of representationLowHigh...Figure 2: General pattern of MFS The sensor units have the responsibility of acquiring the raw data from environment and preprocessing them, and then extracting the information to be fused from the sensordata. A network of fusion units is located above the sensorunits. Information from different sources: sensor units,fusion units and interactive system are combined in each fusion unit. The integration functions play an important rolein MFS. Three functions: sensor selection, world model anddata transformation, can be used to manage and control theprocess. As shown in Fig. 2, the whole MFS is composed ofseveral functional parts and all of the parts can be runconcurrently.2.2 General software framework For MFS, general software framework is described in [6](shown in Fig. 3). It is a three-tier framework that includesman/machine interface, logical part of MFS and database. Man/machine interface is a key area of


View Full Document

UNCC ECGR 6185 - RTOS-based Software Architecture for Multisensor Fusion System

Documents in this Course
Zigbee

Zigbee

33 pages

Load more
Download RTOS-based Software Architecture for Multisensor Fusion System
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view RTOS-based Software Architecture for Multisensor Fusion System and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view RTOS-based Software Architecture for Multisensor Fusion System 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?