NEURAL-NETWORK CONTROL OF A FREE-FLYING SPACE ROBOT

Two recent developments in neutral-network control are presented. Firs t, a ''Fully-Connected Architecture'' (FCA) is developed for Erse with backpropagation (BP). This FCA has functionality beyond that of a lay ered network, and these capabilities are shown to be particularly bene ficial for central tasks. A complexity control method is applied succe ssfully to manage the extra connections provided, and prevent over-fit ting. Second, a technique that extends BP learning to discrete-valued functions (DVFs) is presented. This algorithm is applicable whenever a gradient-based optimization is used for systems with DVFs. The modifi cation to BP is very small simply requiring replacement of the DVFs wi th continuous approximations and injection of noise on the forward swe ep. The viability of both of these neural-network developments is demo nstrated by applying them to a thruster-mapping problem characteristic of space robots. Real-would applicability is shown via an experimenta l demonstration on a 2-D laboratory model of a free-flying space robot .