M. Nuttin, H. Van Brussel, C. Baroglio, and R. Piola. Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In FUZZ-IEEE-94: Third IEEE International Conference on Fuzzy Systems, World Congress on Computational Intelligence, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....techniques can be distinguished, namely fine motion planning and reactive control. The work being done in B Learn II relates to both approaches. Further information concerning fine motion planning can be found in [46, 47] In reactive control, neural networks and fuzzy controllers are employed [17, 19, 18, 21, 40], with a special emphasis on the acquisition of training data from user demonstrations [20, 23] Machining and Monitoring CNC machines play an important role in Flexible Manufacturing Systems: they produce the parts to be assembled. Solving the problem of making products close to the anticipation ....

....itself, i.e. the mapping between sensorial input and the control output to the robot is to be learned. Second, means are necessary to detect the beginning and the end of a phase. Both tasks can be solved by means of function approximation techniques such as neural networks or fuzzy controllers [3, 17, 19, 18, 40]. Apart from knowledge about the structure of the task, the generation of a complete elementary operation requires a description of the set of situations that are valid for the application of this particular operation. Also, the goals that should be achieved by executing the operation must be ....

M. Nuttin, H. Van Brussel, C. Baroglio, and R. Piola. Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In FUZZ-IEEE-94: Third IEEE International Conference on Fuzzy Systems, World Congress on Computational Intelligence, 1994.

....by the tolerance problem. We investigated how much performance degraded when using learnt information on other fixtures. We also showed that it is possible to improve performance when using a new fixture location. The scientific output of this workpackage also can be shown by numerous publications [16, 104, 103, 108, 130, 34, 40, 145, 54, 43, 59, 19, 41, 50, 53, 52, 66, 101, 68, 58, 55, 129, 131, 132, 51, 42, 15, 17], among which several joint publications. 4.3 Results in Assembly and Machining The goal of workpackage WP3 was to enhance supervision architectures in assembly robots and in NC machines by introducing learning capabilities. Activities in the workpackage concerning assembly supervision were ....

M. Nuttin, H. Van Brussel, C. Baroglio, and R. Piola. Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In FUZZ-IEEE-94: Third IEEE International Conference on Fuzzy Systems, World Congress on Computational Intelligence, pages 1217--1223, 1994.

....by the tolerance problem. We investigated how much performance degraded when using learnt information on other fixtures. We also showed that it is possible to improve performance when using a new fixture location. The scientific output of this workpackage also can be shown by numerous publications [16, 104, 103, 108, 130, 34, 40, 145, 54, 43, 59, 19, 41, 50, 53, 52, 66, 101, 68, 58, 55, 129, 131, 132, 51, 42, 15, 17], among which several joint publications. 4.3 Results in Assembly and Machining The goal of workpackage WP3 was to enhance supervision architectures in assembly robots and in NC machines by introducing learning capabilities. Activities in the workpackage concerning assembly supervision were ....

M. Nuttin and C. Baroglio. Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In Proceedings of the Second European Workshop on Learning Robots, Torino, Italy, October 1993.

....is difficult, or even impossible to derive an analytical control function from a quantitative theory, it is easy to find some expert having a qualitative idea of its general shape. In general, this knowledge is the starting point for manual fuzzy controller synthesis (Bonissone and Chiang, 1993; Nuttin et al. 1994). A second source is represented by the ability of a human operator (or of some other machine) of executing the control task him herself. It is extremely hard to explicitly formalize this knowledge as a theory; nevertheless, it is easy to apply it to generate examples of control, and even from bad ....

.... construct also the network layout in an incremental manner are, for instance, GAL (Alpaydin, 1991) and Fritzke s Growing Cell Structures (Fritzke, 1993) The most common approach to design a Fuzzy Controller is to setup the network layout manually, relying on the domain knowledge of a human expert (Nuttin et al. 1994). Variants of the error gradient descent have been proposed in order to refine the fuzzy sets in a second step (Berenji, 1990) In the framework of the present work, three alternative procedures have been defined in order to automate the layout construction of a LRFN. One of the procedures is ....

Nuttin, M., Van Brussel, H., Baroglio, C., and Piola, R. (1994). Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In FUZZ-IEEE-94: Third IEEE International Conference on Fuzzy Systems, World Congress on Computational Intelligence.

....with robots is the automatic generation of assembly plans, which usually contain the actions to be executed from each task state in order to achieve the assembly goal. These states can be defined in different ways (although sometimes it is not explicitly done) such as: using fuzzy set theory [14] [1] performing partitions of the configuration space [13] 5] or classifying the possible contact situation between the objects to be assembled [6] 18] In any case, two main problems arise when uncertainty is considered: a) the identification of the current task state, and b) the automatic ....

.... proposed in order to solve these problems, giving rise to planners that explicitly identify the task state to decide which command must be applied [13] 5] 6] 8] 12] 18] on one hand and, on the other, to reactive systems that directly map sensor information into a robot command [16] 10] 9] [14]. This paper follows the first approach dealing with the problem of current task state estimation in order to follow the assembly strategy determined by a fine motion planner. This work was partially supported by the ESPRIT Project B LEARN II under contract n o 7274, the CICYT project ....

M. Nuttin, H. Van Brussel, C. Baroglio and R. Piola, Fuzzy Controller Synthesis in robotic assembly: Procedure and experiments. FUZZIEEE94: Third International Conference on Fuzzy Systems, World Congress on Computational Intelligence, 1994.

....delayed or scaled input signals, the original definition of the TDNN required that all (delayed) links of a neuron that are connected to one input are identical. This restriction has been weakened for the experiments described here. Further details on Fuzzy Controllers and the TDNN can be found in [47], 32] 5 Synthesis from Examples: an Experiment In order to verify the possibility of synthesizing a controller from examples of correct behaviour, an experiment has been performed using both fuzzy controllers and TDNNs. The test case has been taken from a real robotic application where a ....

M. Nuttin and C. Baroglio. Fuzzy controller synthesis in robotic assembly: Procedure and experiments. In Proceedings of the Second European Workshop on Learning Robots, Torino, Italy, 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