Full neural predictors, with fixed time horizon, for a Truck-Trailer-Trailer prototype of a multi-articulated robot, in backward movements - singular conditions and critical angles

This article comprises a practical and original application of full neural predictors with fixed prediction horizon in backward movements of a Truck-Trailer-Trailer prototype of a multi-articulated mobile robot (MAMR), in the configuration space. It´s used a new proposal based on static multilayer feedforward networks. This kind of predictor is useful for assisted operations or can be used as cores in simulators to analyze navigation strategies and for controller´s synthesis and validation. The systematic and the presented tools are general. The training data set is composed by real data acquired from measurements of the prototype and by data generated from singular condition models. The article uses original models for the singularities and for the critical angles. These models were deduced from general movement equations of a MAMR with on-axle or off-axle hitching and with front or rear traction on the truck. The use of models for singularities is necessary because the circular conditions are situations of unstable equilibrium, which makes impossible to obtain enough data from open loop real systems. The model for critical angles is used to define the range for data acquisition before the jackknife. A MAMR´s prototype is used for data acquisition and for synthesis and validation of neural networks. The characteristics of this robot are also presented. It is shown the results of the procedures applied to the collected data and to predictors with different prediction horizons during their training and validation, using the created Interface. The results demonstrate the good performance of the systematic and tools.