McCarragher, B. J. and Asada, H., 1993, Qualitative template matching using dynamic process models for state transition recognition of robotic assembly, Transactions of the ASME, Journal of Dynamic Systems, Measurement, and Control, Vol. 115, pp. 261--269.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of sensing data sets required and the number of equations. However, they do assume that the contact conditions remain unchanged during the identification process (i.e. no slip occurs) McCarragher and Asada used an analysis of rigid body dynamics to identify the contact formation between objects [44]. Constraints were added to the equations of motion with Lagrange multipliers and a velocity constraint matrix. For each discrete contact formation, all possible quality states were enumerated, where a quality state is defined as a unique combination of positive, negative, or zero values for ....

B. J. McCarragher and H. Asada, "Qualitative template matching using dynamic process models for state transition recognition of robotic assembly," In Modeling and Control of Compliant and Rigid Motion Systems, volume 31, pp. 155--163, Atlanta, GA, December 1991, ASME.

....descriptions of the evolution of the assembly have been developped. In [25] in order to help the monitoring of the sequence of hybrid control primitives, a qualitative description of the expected evolution of the object velocity and the contact forces is associated with the assembly program. In [31], a qualitative representation of the object dynamic is used to recognize in a very fast way when a particular contact state transition takes plave. This dynamic approach to the detection of transitions is different from the two more classical ones, which consist in detecting contact relations ....

B. J. McCarragher and H. Asada. Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotics Assembly. Journal of Dynamic Systems, Measurement, and Control, 115, 1993.

....requires very structured and hence expensive setups if one requires absolute certainty about the task sequence. Loosening the requirements for the set up introduces increased uncertainty about the sequence of contact situations. This problem has been tackled using a Hidden Markov Model approach, [8, 12, 18]: the di#erent contact situations correspond to states in the network (each with a typical set of force motion 0 100 100 10 100 0 1.0 Figure 5: Example of a template to be used in a Monte Carlo localization algorithm for finding the planar top of a cylindrical barrel, with a ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Trans. ASME J. Dyn. Systems Meas. Control, 115:261--269, 1993.

....they solved a linear program to determine CF feasibility, and ranked feasibilityby the distance between the measured force and the cone of allowable forces. However, solving the linear program was found to be too slow for real time. McCarragher and Asada used rigid body dynamics to identify CFs [8]. Constraints were added with Lagrange multipliers and a velocity constraint matrix. Qualitative states (QS defined as a unique combination of positive, negative, or zero values for position, velocity, and acceleration) were enumerated. Templates, precalculated as a sequence of QSs for contact ....

B. J. McCarragher and H. Asada, "Qualitative template matching using dynamic process models for state transition recognition of robotic assembly," Journal of Dynamic Systems Measurement and Control-- Transactions of the ASME,vol. 115, no. 2, pp. 261--269, June 1993.

....requires very structured and hence expensive setups if one requires absolute certainty about the task sequence. Loosening the requirements for the set up introduces increased uncertainty about the sequence of contact situations. This problem has been tackled using a Hidden Markov Model approach, [8, 12, 18]: the di erent contact situations correspond to states in the network (each with a typical set of force motion 0 100 100 10 100 0 1.0 Figure 5: Example of a template to be used in a Monte Carlo localization algorithm for nding the planar top of a cylindrical barrel, with a Gaussian like ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Trans. ASME J. Dyn. Systems Meas. Control, 115:261-269, 1993.

....alarms from which it is difficult to recover. Using more detailed sensor models (including the stochastic properties of the sensors as well as a more detailed and flexible geometric model of the local contact geometry) some promising results to improve the reliability have been obtained already, [5, 8, 14, 15, 6]. 4.3 Non polyhedral objects Following non polyhedral surfaces is generally not a big deal for a force controlled robot. However, it s much more difficult to keep track accurately of where one is moving over the surface, 3] so that on line and two way synchronisation between planner and robot ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Trans. ASME J. Dyn. Systems Meas. Control, 115:261--269, 1993.

....of active compliant techniques can be distinguished: fine motion planning and reactive control, and it is in a reactive control approach that we have integrated reinforcement learning techniques and neural nets. More information on compliant motion and the peg into hole task can be found in [3, 7, 10, 9, 14, 15]. 2 The reinforcement learning (RL) system 2.1 Introduction Reinforcement learning (RL) is an universal applicable but slowly converging optimisation technique [8, 11, 16] Typically RL is applied when 1) a model isn t available, and 2) only a scalar evaluation of the true performance is ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Journal of Dynamic Systems, Measurement, and Control, pages 261--269, June 1993.

....implemented systems yet, 31, 32] Explicitly sensor based actions. De Schutter and Van Brussel [33] introduce compliant moves that perform tracking of orientation uncertainties in the TF. Yoshikawa and Sudou [34] integrate this into the dynamic control of the manipulator. McCarragher and Asada [35], and Eberman and Salisbury [3] develop on line state transition detection, to guide the robot through a set of prede ned states that can occur in 2D contact problems. Wampler [36] and Samson et al. 10] describe some elementary compliant actions , with other than force sensors. On line contact ....

B. J. McCarragher and H. Asada, \Qualitative template matching using dynamic process models for state transition recognition of robotic assembly," Trans. ASME, J. Dyn. Syst., Meas., and Control, vol. 115, pp. 261-269, 1993.

....algorithms that we want to integrate learning techniques. Two classes of active compliant techniques can be distinguished: fine motion planning and reactive control. This paper follows the reactive control approach. More information on compliant motion and the peg into hole task can be found in [9, 11, 15, 16, 23]. 1 This section can be skipped by readers only interested in fuzzy controller synthesis. 3 A fuzzy controller for the peg into hole task 3.1 The fuzzy controller The strength of a fuzzy controller is its capability to handle both linguistic knowledge and numerical sensor data. For the ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Journal of Dynamic Systems, Measurement, and Control, pages 261--269, June 1993.

....Decoder (Defuzzifier) Figure 1: basic architecture of a fuzzy controller. techniques can be distinguished: fine motion planning and reactive control. And it is a reactive control algorithm that we want to synthesise. More information on compliant motion and the peg into hole task can be found in [8, 11, 14, 13, 20]. 3 A fuzzy controller for the peg into hole task The strength of a fuzzy controller (FC) is its capability to handle both linguistic knowledge and numerical sensor data. For the presented application, fuzzy control is also a means to achieve non linear control. The basic fuzzy control ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Journal of Dynamic Systems, Measurement, and Control, pages 261--269, June 1993.

....of the observer z(t) associated with a pro cess state transition are formulated as a set of rules which are then used to recognise the corresponding transition. Rule based process monitoring overcomes several problems associated with approaches based on template matching and qualitative reasoning [21]. The rules are more general than templates which are based on the current value of the observer. Also, the rules are obtained by demonstration rather than qualitative reasoning methods which have the problem that consistent qualitative algebras are difficult to obtain [36] The organisation of ....

B.J. McCarragher and H. Asada. "Qualitative template matching using dynamic process models for state transition recognition of robotic assembly", The ASME Journal of Dynamic systems, Measurement and Control, 115(2A):261-275, June 1993.

.... dynamics of the task, as opposed to previous work in assembly based on the quasi static assumption which ignores dynamic effects (see, for example, Whitney, 1982; Bruyninckx et al. 1995 ] Our approach to obtaining the rules from examples can be contrasted with the model based approach in [ McCarragher and Asada, 1993 ] where qualitative reasoning techniques are used to obtain simple templates for reconizing contact state transitions. There is only one template for each contact state transition. The templates are based on the current values of quantized variables and do not consider temporal sequences. ....

Brenan J. McCarragher and Haruhiko Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. The ASME Journal of Dynamic systems, Measurement and Control, 115(2A):261--275, June 1993.

....between the workpiece and the environment. The possible contact formations are then the formations where the measured forces moments are contained in this set. This method only applies to systems with quasi static motions, negligible friction and noiseless sensing. A different approach is used by [McCarragher and Asada, 1993]. Assuming polygonal models of the workpiece and the environment, they highlight the discrete changes of state. A discrete change of state is defined as a change in the geometric constraint between the workpiece and the environment. This method uses qualitative force torque measurements and ....

....only. The simplicity of the filters in the time domain, H x (t) and H y (t) is an advantage in fast on line event detection. The event detector used here for a planar assembly task is described by the following threshold test. H x (t) 2 H y (t) 2 ( DeltaF ) 2 (17) As in the paper by [McCarragher and Asada, 1993], we determine the threshold by looking at the standard deviation of the measurements. We do not want the effects from the friction forces to trigger the threshold test, so the workpiece was slid across a surface and ( DeltaF ) 2 was chosen to be a 3oe deviation from the mean value of H x (t) ....

B.J. McCarragher and H. Asada. 1993. Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotic Assembly. ASME Journal of Dynamic Systems, Measurements and Control, Vol. 115, no. 2A, pp.261-269.

....importantly, the approach can be adapted quickly to different tasks by simply learning a new set of discriminant functions from sensory data corresponding to the task. Since the learning phase is based on a demonstration of the task, our approach is applicable both to robotic process monitoring [9] and to the study of human demonstration of tasks [8] Finally, our approach is general in that it can be applied to one or more sensory modalities such as position and force. 2 System Modeling We model robotic tasks as hybrid dynamic systems. A discrete event system is used to describe the task ....

....enough that dynamic effects can be ignored (see, for example, 11, 4] However, the movements used to carry out the task occur in the continuous domain and the assembly process is dynamic in nature, especially during changes in the discrete states of contact. Recently, McCarragher and Asada [9] focused on the dynamic nature of the assembly process and presented a new technique for real time process monitoring based on the notion of qualitative reasoning [3] The process monitor used in this study is based on qualitative reasoning involving a discretized force signal. Each state and ....

[Article contains additional citation context not shown here]

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. The ASME Journal of Dynamic systems, Measurement and Control, 115(2A):261-- 275, June 1993.

....force. Figure 2: A part of the contact state network for the peg in hole task. Each circle denotes a contact state, with the figure inside the circle illustrating the corresponding peg in hole configuration. Each arc represents an allowed change in contact state. Recently, McCarragher and Asada [4] focused on the dynamic nature of the assembly process and presented a new technique for real time process monitoring based on the notion of qualitative reasoning [2] In the approach adopted in [4] the continuous force signals are discretized into a small number of discrete qualitative signals, ....

....Each arc represents an allowed change in contact state. Recently, McCarragher and Asada [4] focused on the dynamic nature of the assembly process and presented a new technique for real time process monitoring based on the notion of qualitative reasoning [2] In the approach adopted in [4], the continuous force signals are discretized into a small number of discrete qualitative signals, which are then used to make decisions about state transitions based on a template matching procedure. The templates are obtained from the discretized version of the process dynamic equations using ....

[Article contains additional citation context not shown here]

Brenan J. McCarragher and Haruhiko Asada. Qualitative template matching using dynamic process models for state transition recopnition of robotic assembly. The ASME Journal of Dynamic systems, Measurement and Control, 115(2A):261--275, June 1993.

....using the distance functions. In most practical situations accurate models are not available and the output of the monitor may be erroneous. The main advantage of the method is a low computational effort. Process Monitor 2 is based on force measurements and qualitative template matching, [6]. The forces resulting from interaction between the workpiece and the environment contain a significant amount of information about the assembly contact state. The template matching monitor exploits this fact by analysing the qualitative changes DeltaF x and DeltaF y of the force signal. From a ....

B.J. McCarragher and H. Asada, Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotic Assembly. ASME Journal of Dynamic Systems, Measurements and Control, Vol. 115, no. 2A, 1993, pp.261-269.

....Information from different sensors are combined into one single outcome, with (hopefully) better resolution and reliability than could be produced by the single sensors individually. 4. Classification, also known as (template pattern) matching, labelling or event object pattern recognition, [10, 22, 21, 20, 35, 49, 74, 83, 79]. A (spatial or temporal) sequence of sensor signals is compared to logs of previous executions, or to models of the task, in order to find the best correspondence. 5. Sequence planning, or decision making, 2, 1, 14, 17] If the task controller has the option to choose between different ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Trans. ASME J. Dyn. Systems Meas. Control, 115:261--269, 1993.

.... c i (t) DES e x v (t) i Figure 1: A Hybrid Dynamic System the robot manipulator and its controller have been designed using the available modelling and control schemes (see [1] 18] The same assumption is applied for the first function of interface by using the process monitoring proposed in [14]. The discrete event model of the manipulation task is defined according to contact states and changes in contact. A manipulation DES is defined as a finite directed graph G = M; E) composed of a finite set of discrete states M = ffl 1 ; fl 2 ; fl m g and a finite set of discrete events ....

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognation of robotic assembly. ASME Journal of Dynamic Systhems Measurement and Control, 115(2A):pp. 261--269, June 1993.

....for compliant motion assemblies is described. The theory provides a technology for constructing plans that might work, but fail in a reasonable way when they cannot. However, the direct applicability of this theory is limited. Another example is found in the paper by McCarragher and Asada, [2]. They present a model based approach which incorporates the dynamic nature of the process to highlight the discrete changes of state. This method uses qualitative force torque measurements and hence is well suited for fast real time event detection. However, by using qualitative measurements, a ....

B.J. McCarragher and H. Asada, Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotic Assembly. Journal of Dynamic Systems, Measurements and Control, June 1993, Vol. 115, no. 2A, p.261-269.

....of the observer z(t) associated with a process state transition are formulated as a set of rules which are then used to recognise the corresponding transition. Rule based process monitoring overcomes several problems associated with approaches based on template matching and qualitative reasoning [21]. The rules are more general than templates which are based on the current value of the observer. Also, the rules are obtained by demonstration rather than qualitative reasoning methods which have the problem that consistent qualitative algebras are difficult to obtain [36] The organisation of ....

B.J. McCarragher and H. Asada. "Qualitative template matching using dynamic process models for state transition recognition of robotic assembly", The ASME Journal of Dynamic systems, Measurement and Control, 115(2A):261--275, June 1993.

....for compliant motion assemblies is described. The theory provides a technology for constructing plans that might work, but fail in a reasonable way when they cannot. However, the direct applicability in robotics is limited. Another example is found in the paper by McCarragher and Asada, [7]. They present a model based approach which incorporates the dynamic nature of the process to highlight the discrete changes of state. This method uses qualitative force torque measurements and hence is well suited for fast real time monitoring in robotic assembly. However, by using qualitative ....

B.J. McCarragher and H. Asada, Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotic Assembly. Journal of Dynamic Systems, Measurements and Control, June 1993, Vol. 115, no. 2A, p.261-269.

....for compliant motion assemblies is described. The theory provides a technology for constructing plans that might work, but fail in a reasonable way when they cannot. However, the direct applicability in robotics is limited. Another example is found in the paper by McCarragher and Asada, [2]. They present a model based approach which incorporates the dynamic nature of the process to highlight the discrete changes of state. This method uses qualitative force torque measurements and hence is well suited for fast real time monitoring in robotic assembly. However, by using qualitative ....

....of the filters in the time domain, H x (t) and H y (t) is an advantage in fast on line event detection. The event detector used here for a planar assembly task is described by the following threshold test. H x (t) 2 H y (t) 2 DeltaF (6) As in the paper by McCarragher and Asada [2], we determine the threshold by looking at the standard deviation of the measurements. We do not want the effects from the friction forces to trigger the threshold test, so the workpiece was slid across a surface and DeltaF was chosen to be a 3oe deviation from the mean value of H x (t) 2 H y ....

B.J. McCarragher and H. Asada, Qualitative Template Matching Using Dynamic Process Models for State Transition Recognition of Robotic Assembly. Journal of Dynamic Systems, Measurements and Control, June 1993, Vol. 115, no. 2A, p.261-269.

No context found.

McCarragher, B. J. and Asada, H., 1993, Qualitative template matching using dynamic process models for state transition recognition of robotic assembly, Transactions of the ASME, Journal of Dynamic Systems, Measurement, and Control, Vol. 115, pp. 261--269.

No context found.

McCarragher, B. J. and Asada, H., 1993, Qualitative template matching using dynamic process models for state transition recognition of robotic assembly, Transactions of the ASME, Journal of Dynamic Systems, Measurement, and Control, Vol. 115, pp. 261--269.

No context found.

B. J. McCarragher and H. Asada. Qualitative template matching using dynamic process models for state transition recognition of robotic assembly. Trans. ASME J. Dyn. Systems Meas. Control, 115:261--269, 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC