We describe a genetic algorithm approach to the automated design of vibratory bowl part feeders. Our approach gives us near-optimal designs in much less time than previously published optimal, brute-force search methods. We have implemented our approach in an automated part feeder design system, and we present preliminary results generated by our system.

We present a hybrid approach to optimize the counterweight balancing of a robot arm, which uses a combination of a genetic algorithm (GA) with the min-max multiobjective optimization method to get the Pareto optimal set of solutions. This set corresponds to several possible robot designs from which the most appropriate has to be chosen by the designer. Our approach is compared to a more traditional min-max search technique in which a combination of random and sequential search was used to generate the Pareto optimal solutions. Our results show how the GA is able to get solutions with a lower deviation from the ideal vector.

The paper presents a method for optimizing the design of plane and space trusses subject to a specified set of constraints. The method is based upon a search technique using genetic algorithms. Traditional structural optimization techniques consider it continuous search space, and consequently lead to unrealistic solutions because structural members are not available in continuously varying sizes. A practical method should consider only the discrete values associated with commonly available materials. On the other hand, most modern structural optimization techniques, even when they consider a discrete search space, suffer a lack of generality, and tend to be limited to a certain kind of structure. Genetic algorithms remedy these two problems since they can deal with discrete search spaces and they are general enough to be easily extended to any kind of structure without substantial modifications. Our results show the genetic algorithm can provide very good solutions, often surpassing other complex and specialized techniques.>

A method of computing manipulation plans is introduced. The empirical backprojection method uses observed state transitions to derive predictions of action effects and links these predictions to form manipulation plans. The data-intensive approach is in contrast to manipulation planning approaches that uses analysis of task mechanics to generate predictions. The empirical backprojection method is described, and three example tasks are used to illustrate the method. The assumptions underlying the method are then compared to the assumptions required by previously published analytical techniques.>

A description is given of a robot, possessing limited sensory and effectory capabilities but no initial model of the effects of its actions on the world, that acquires such a model through exploration, practice, and observation. By acquiring an increasingly correct model of its actions, it generates increasingly successful plans to achieve its goals. In an apparently nondeterministic world, achieving reliability requires the identification of reliable actions and a preference for using such actions. Furthermore, by selecting its training actions carefully, the robot can significantly improve its learning rate.>