A lightweight slender robot concept could also be useful fortasks in the final assembly lines of automobile plants。 Thescenario in this case must, however, also include the possibilityto make intuitive programming and efficient failure handling indirect contact between human and robot, otherwise theFig。 7。 Capacitive encoder rings (Hexagon Metrology) integrated into an ABBrobot joint for accurate measurements of robot arm movements。 automation systemwill be too complex。 Therefore, full safety isneeded and a low moving mass and compliant control of therobot is important in this respect (DLR, 2006; Hirzinger, Albu-Scha ¨ffer, Ha ¨hnle, Schaefer, & Sporer, 2001; Ogawa, Haniya,Okahisa, & Ichibangase, 2005; Zinn, Khatib, Roth, &Salisbury, 2004)。 In order to obtain a high safety level, it isnot far away to make use of the sensors needed for virtualstiffness control also for safety purpose。 This could be done byusing the redundant measurement signals to generate residualsand to make redundant supervision of the robot control。 If also asix DOF force/torque sensor or joint torque sensors are used forhuman–robot interaction, further redundant supervision can beintroduced。A robot concept for flexible automation of the final assemblyof cars could also be used for the assembly of other complexhigh volume products, to be found, for example, in the homeappliancemanufacturing industries。 It could also be used for theautomation of assembly of products with smaller lot sizes andthen the concept has some resemblance with the robot assistantconcepts found in academic research today (Ha ¨gele, Schaaf, &Helms, 2002)。 However, in order to be able to obtain aneconomically feasible solution the robots in the mentionedassembly applications cannot work as an assistant to theoperator but instead the operator should assist the robots tosolve unpredictable problems。 The robot installations shouldthen be so robust that one operator could serve several robots。 Inthis respect, it is very important to have a good balance betweenthe autonomy of the robot control and the tasks for the operator。For example, the robot could have some basic failure recoveryschemes but infrequent difficult to handle failure situationsshould be handled by the operator by direct interaction with therobot。 An interesting aspect of developing robot control forsuch human robot collaboration is that the software architecturein industrial robot controllers of today will not be optimal andneither the software architectures for autonomous systems。Thus, new efficient control architectures are needed and the useof a combination of industrial and academic experiences isprobably necessary to develop high performance scalablesoftware architectures for these new robot concepts。A robot that is able to work automatically according to aprogram generated, supervised, edited and recovered by directphysical interaction between human and robot will of coursealso solve many of the problems encountered when introducingrobots in small and medium size enterprises (SMErobot, 2005)。However, these companies usually have much smaller lot sizesthan the automotive industry and they represent a much largerspectrum of applications and automation environments thanfound in the automotive final assembly lines。 Moreover, theireconomy is much more sensitive to big investments and they donot have the access to the automation and robot expertise foundin the organisation of an automotive manufacturer。
Thequestion then is if it is possible to develop a robot family atthe necessary price level for all the small and medium sizedenterprises (SME) applications and requirements。 One prerequisite for success in this case could be to find a modularrobot program, which means that industrial robots will bepossible to assemble for different work space sizes, for differentload weights and for different performance requirements。 Theimplication of this is that also the model-based robot controlmust be modular, including modular kinematics and dynamicmodels。 To obtain optimal control for all possible moduleconfigurations, automatic servo- andmodel tuning is needed foreach configuration of the robot modules。 Since such a highlymodular robot concept will increase the difficulties atconfiguration, re-configuration and maintenance, it will beimportant to hide the complexity for the user。 This means thatfunctionalities must be developed for Plug and Play andintuitive human–robot interaction throughout the lifecycle ofthe robots。 What this could imply is that robot CAD systemsneed to support robot cell commissioning and design usingvirtual environment with configuration tools handling Plug andPlay components for robot- and cell modules (Gauss, Dai,Zimmermann, Som, & Wo ¨rn, 2006)。 When the selectedmodules are delivered to the SME site the robot controllershould perform automatic cell configuration supported by Plugand Play to make the installation of mechanical, electrical andsoftware modules very simple。 At a re-installation of the robotin another application the same configuration tools can be usedagain, making it easy for an SME to adapt the robot-basedproduction to new products and new customer requirements。 Inthis scenario, new concepts are also needed to be able to supporteasy to use intuitive robot- and cell calibration and robotprogramming。 One component in such a solution could be theuse of multi modal robot–human interaction including leadthrough (Fig。 8) for the positioning of the robot。One interesting possibility when building a modular robotfamily is to use parallel kinematics。 Actually, this could berealistic using recently reported parallel kinematic structureswith a small footprint in relation to the work space as for serialrobots (Fig。 9; Cui, Zhu, Gan, & Broga ˚rdh, 2005; Williams,Hovland, & Broga ˚rdh, 2006)。 Since the six parallel links inthese new structures only need to transmit axial forces and nobending or twisting torques, it is easy to build lightweightrobots with very high accuracy and stiffness。 These propertiesare important in, for example, the fettling applicationsmentioned earlier and since high performance combined withlow inertia can easily be obtained also for a large work space,the new structures will be useful, for example, for theprocessing and assembly of aerospace components and for theprocessing of other large construction parts as for wind mills,bridges, buildings, railroads, etc。, as mentioned before。Moreover, because the mechanical robot structure is notredundant, it is very easy to assemble this type of robots withoutany requirements on mounting accuracy。 This makes thesestructures ideal for Plug and Play and it could even be realisticto assemble and re-assemble the robots at the customer sitewithout any advanced mounting equipment。In future robot applications with very complex work objectsas in disassembly, sorting, cleaning, cutting up etc。 the robotsmust be supported by advanced vision systems。 Probably, 3Dcombined with colour and texture recordings (Suppa &Hirzinger, 2005) is needed for robust recognition- andmeasurements on the object structures that will be processedand the vision systems could need assistance from other sensor types to measure not visible characteristics of the work objects。Because of the huge variability in object features andprocessing rules efficient data bases must be built to modelthe relations between object characteristics and robot tasks。These process models should be possible to build using thecraftsmen’s experience。 未来工业机器人控制的发展英文文献和中文翻译(6):http://www.751com.cn/fanyi/lunwen_62458.html