Real-Time Visual Grasp Synthesis Using Genetic Algorithms and Neural Networks

This paper addresses the problem of automatic grasp synthesis of unknown planar objects. In other words, we must compute points on the object’s boundary to be reached by the robotic fingers such that the resulting grasp, among infinite possibilities, optimizes some given criteria. Objects to be grasped are represented as superellipses, a family of deformable 2D parametric functions. They can model a large variety of shapes occurring often in practice by changing a small number of parameters. The space of possible grasp configurations is analyzed using genetic algorithms. Several quality criteria from existing literature together with kinematical and mechanical considerations are considered. However, genetic algorithms are not suitable to applications where time is a critical issue. In order to achieve real-time characteristics of the algorithm, neural networks are used: a huge training-set is collected off-line using genetic algorithms, and a feedforward network is trained on these values. We will demonstrate the usefulness of this approach in the process of grasp synthesis, and show the results achieved on an anthropomorphic arm/hand robot.