APPLICATIONS OF NEURAL NETWORKS TO ROBOTICS

Real-time and intelligent processing in new situations are requirements imposed by the robotics field that are impossible for only barely possible) to be carried out using today's analytical methods. In the following, neural networks that seem predestined for application in robotics on the basis of their operating characteristics are introduced. After a short introduction, the two types of networks most applied, backpropagation and Hopfield networks, are described in detail. Finally, a synopsis of neural network solution approaches in the robotics field is made, with a closer look at solutions to kinematic and dynamic problems, as well as interesting attempts at sensor evaluation and trajectory planning.