Abstract
This paper presents an external wrench estimator that uses a hybrid dynamicsmodel consisting of a first-principles model and a neural network. Thisframework addresses one of the limitations of the state-of-the-art model-basedwrench observers: the wrench estimation of these observers comprises theexternal wrench (e.g. collision, physical interaction, wind); in addition toresidual wrench (e.g. model parameters uncertainty or unmodeled dynamics). Thisis a problem if these wrench estimations are to be used as wrench feedback to aforce controller, for example. In the proposed framework, a neural network iscombined with a first-principles model to estimate the residual dynamicsarising from unmodeled dynamics and parameters uncertainties, then, the hybridtrained model is used to estimate the external wrench, leading to a wrenchestimation that has smaller contributions from the residual dynamics, andaffected more by the external wrench. This method is validated with numericalsimulations of an aerial robot in different flying scenarios and differenttypes of residual dynamics, and the statistical analysis of the results showsthat the wrench estimation error has improved significantly compared to amodel-based wrench observer using only a first-principles model.