关 键 词：

大学生 方程式赛车 设计 总体 三维

资源描述：

喜欢这套资料就充值下载吧。资源目录里展示的都可在线预览哦。下载后都有，请放心下载，文件全都包含在内，图纸为CAD格式可编辑，有疑问咨询QQ:414951605 或 1304139763

内容简介：

The MIT Indoor Multi-Vehicle Flight TestbedMario Valenti, Brett Bethke, Daniel Dale, Adrian Frank,James McGrew, Spencer Ahrens, Jonathan P. How, and John VianAbstractThis paper and video present the componentsand flight tests of an indoor, multi-vehicle testbed that wasdeveloped to study long duration UAV missions in a controlledenvironment. This testbed is designed to use real hardware toexamine research questions related to single- and multi-vehiclehealth management, such as vehicle failures, refueling, andmaintenance. The testbed has both aerial and ground vehiclesthat operate autonomously in a large, indoor flight test area andcan be used to execute many different mission scenarios. Thesuccess of this testbed is largely related to our choice of vehicles,sensors, and the systems command and control architecture.The video presents flight test results from single- and multi-vehicle experiments over the past year.BACKGROUNDUnmanned vehicles are being used by a number of or-ganizations to locate, observe and assess objects of interestfrom sophisticated operator stations miles from the area ofoperations. While many researchers have been discussingautonomous multi-agent operations 1, 2, more work isneeded on how to perform health management for au-tonomous task groups. In the past, the term “health manage-ment” was used to define systems which actively monitoredand managed vehicle sub-systems (e.g., flight controls, fuelmanagement, avionics) in the event of component failures.In the context of multiple vehicle operations, this definitioncan be extended to autonomous multi-agent groups: teamsinvolved in a mission serve as a “vehicle system,” eachvehicle is a sub-system of each multi-agent team, and soon.As discussed in 3, many research groups have usedoutdoor test platforms to verify theories relating to innovativeIEEE Member, Ph.D. candidate, Department of Electrical Engineeringand Computer Science, Massachusetts Institute of Technology, Cambridge,MA 02139valentiIEEE Student Member, S.M. Candidate, Department of Aeronautics andAstronautics, Massachusetts Institute of Technology, Cambridge, MA 02139bbethkeM.Eng. Candidate, Department of Electrical Engineering and ComputerScience, Massachusetts Institute of Technology, Cambridge, MA 02139ddaleVisitingStudent,DepartmentofAeronauticsandAstronautics,MassachusettsInstituteofTechnology,Cambridge,MA02139a frankS.M.Candidate,DepartmentofAeronauticsandAstronautics,MassachusettsInstituteofTechnology,Cambridge,MA02139jsmcgrewS.M. Candidate, Department of Mechanical Engineering, MassachusettsInstitute of Technology, Cambridge, MA 02139sahrensAssociate Professor of Aeronautics and Astronautics, Aerospace ControlsLaboratory, Massachusetts Institute of Technology, 77 Massachusetts Ave,M.I.T. Rm 33-326, Cambridge, MA 02139jhowAuthor towhom all correspondence should be addressedTechnicalFellow,BoeingPhantomWorks,Seattle,WA98124john.vianUAV concepts 4, 5, 6, 7. In addition, there are anumber of indoor testbeds that have been built to studymulti-agent activities 8, 9. These testbeds have a numberof limitations that inhibit their utility for investigating thehealth management of UAV teams performing large-scalemissions over extended periods of time. For example, out-door platforms can only be tested during the proper weatherand environmental conditions. In addition, these externalUAVs also typically require a large support team, whichmakes long-term testing logistically difficult and extremelyexpensive. Furthermore, many of the indoor testbeds areeither 2D or operate in 3D in a very limited flight volume.In contrast, the MIT indoor multi-vehicle testbed isuniquely designed to study long duration missions in a con-trolled environment. This testbed is being used to implementand analyze the performance of techniques for embeddingthe fleet and vehicle health state into the mission and UAVplanning. In particular, we are examining research questionsrelated to vehicle and multi-agent health management issues,such as vehicle failures, refueling and maintenance usingreal hardware. The testbed is comprised of aerial and groundvehicle components, allowing researchers to conduct tests fora wide variety of mission scenarios. We have demonstrateda number of multi-vehicle coordinated test flights (usingboth autonomous ground and air vehicles). Currently, we areable to fly more than four air vehicles in a typical-sizedroom, and it takes no more than one operator to set up theplatform for flight testing at any time of day for any lengthof time. At the heart of the testbed is a global metrologysystem that yields very accurate, high bandwidth positionand attitude data for all vehicles in the entire room. Ourtestbed configuration does not require modifications to off-the-shelf radio-controlled vehicle hardware. As a result, thisplatform is ideal for the rapid prototyping of multi-vehiclemission management algorithms since it can be operated overlong periods of time using one person at a fraction of thecost of what would be needed to support an external flightdemonstration.TESTBEDARCHITECTURE ANDCOMPONENTSAs shown in Figure 1, the testbed architecture has fourmajor components; a mission planning level to set the systemgoals and monitor system progress, a task assignment levelwhich (in general) assigns specific tasks to a vehicle orvehicle group in order to support the overall mission goals,a trajectory design level which directs each vehicle and itssubsystems on how to best perform the actual tasks providedby the task processing level, and a control processing level2007 IEEE International Conference onRobotics and AutomationRoma, Italy, 10-14 April 2007ThC12.11-4244-0602-1/07/$20.00 2007 IEEE.2758Figure 1: Testbed Architecture Block Diagramdesigned to carry out the activities set by upper-levels in thesystem. Note that health information about each componentin the system is provided to and used by each componentin the architecture in the decision making process. Thehealth monitoring components are designed to evaluate theperformance of each component and provide feedback to therest of the system (and operator) regarding its progress andmission effectiveness. This information is used by the rest ofthe system to potentially adjust or redirect its mission and/ortask goals as the mission is taking place.To test and demonstrate the real-time capabilities of thesehealth management algorithms in a realistic, real-time envi-ronment, we developed a low-cost, indoor testing environ-ment which could be used over extended periods of time ina controlled environment. The vehicles used in the platformare commercially available and off-the-shelf (COTS) R/Cvehicles is designed to be durable and safe making themsuitable for an indoor flight testbed. In addition, no structuralor electronics modifications were made to the air vehiclesused in these flight tests, which helps to maximize flighttime and reduce stress on the motor/blade components. Allcomputing for this system is done on ground-based comput-ers, which have two AMD 64-bit Opteron processors, 2 Gbof memory and run Gentoo Linux. The control and commandprocessing is processed by this computer and sent over anRS-232 connection from the vehicles control computer tothe vehicles R/C Transmitter over the transmitters trainerport interface.A Vicon MX camera system 10 is used to detect thevehicles position and orientation in real-time. By attachinglightweight reflective balls to the vehicles structure, theVicon MX Camera system can track and compute eachvehicles position and orientation information up to 120 Hzwith a 10 ms delay. This data is then transmitted via ethernetto each vehicles ground based control computer.In addition, the testbed is designed with an automatedsystem task manager. Since each air vehicle in the systemcan take-off and land autonomously, the task manager au-tonomously manages every air and ground vehicle controlledin the system using these task level commands (shownin Figure 2). As a result, multi-vehicle mission scenarios(e.g., search, persistent surveillance) can be organized andimplemented by the task manager autonomously. Finally, thesystem is designed to allow an operator to issue a commandto any vehicle (at any time) through the operator interface,which includes a 3D display of the objects in the testing areaand a GUI that displays vehicle health and state data, taskinformation, and other mission-relevant data.Figure 2: Testbed Command & Control ArchitectureFor more information and papers about this researchproject, visit ACKNOWLEDGEMENTSThe authors would like to thank Spencer Ahrens, GlennTournier and Michael Robbins for their invaluable assistancein the project. Brett Bethke would like to thank the HertzFoundation and the American Society for Engineering Edu-cation for their support of this research. This research hasbeen generously supported by Boeing Phantom Works inSeattle, WA. Research also funded in part by AFOSR grantFA9550-04-1-0458.REFERENCES1 P. Gaudiano, B. Shargel, E. Bonabeau, and B. Clough, “Control ofUAV SWARMS: What the Bugs Can Teach Us,” in Proceedingsof the 2nd AIAA Unmanned Unlimited Systems, Technologies, andOperations Aerospace Conference, San Diego, CA, September 2003.2 H. Paruanak, S. Brueckner, and J. Odell, “Swarming Coordination ofMultiple UAVs for Collaborative Sensing,” in Proceedings of the 2ndAIAA Unmanned Unlimited Systems, Technologies, and OperationsAerospace Conference, San Diego, CA, September 2003.3 M. Valenti, B. Bethke, G. Fiore, J. How, and E. Feron, “Indoor Multi-Vehicle Flight Testbed for Fault Detection, Isolation, and Recovery,”in Proceedings of the AIAA Guidance, Navigation, and Control Con-ference and Exhibit, Keystone, CO, August 2006.4 D. Shim, H. Chung, H. J. Kim, and S. Sastry, “Autonomous Ex-ploration in Unknown Urban Environments for Unmanned AerialVehicles,” in Proceedings of the AIAA Guidance, Na