We introduce GP-FNARX: a new model for non linear system identification based on a nonlinear autoregressive exogenous model (NARX) with filtered regressors (F) where the nonlinear regression problem is tackled using sparse Gaussian processes (GP). We integrate data pre-processing with system identification into a fully automated procedure that goes from raw data to an identified model. Both pre-processing param eters and GP hyper-parameters are tuned by maximizing the marginal likelihood of the probabilistic model. We obtain a Bayesian model of the system's dynamics which is able to report its uncertainty in regions where the data is scarce.The automated approach, the modeling of uncertainty and its relatively low computational cost make of GP-FNARX a good candidate for applications in robotics and adaptive control.

Gaussian process state-space models (GP-SSMs) are a very flexible family of models of nonlinear dynamical systems. They comprise a Bayesian nonparametric representation of the dynamics of the system and additional (hyper-)parameters governing the properties of this nonparametric representation. The Bayesian formalism enables systematic reasoning about the uncertainty in the system dynamics. We present an approach to maximum likelihood identification of the parameters in GP-SSMs, while retaining the full nonparametric description of the dynamics. The method is based on a stochastic approximation version of the EM algorithm that employs recent developments in particle Markov chain Monte Carlo for efficient identification.

Abstract-Many different robotic in-parallel structures have been conceived as six-component force sensors. In general, they perform well for most applications but, when accuracy is a must, two main limitations arise. First, in most designs, the legs are connected to the base and the platform through balland-socket joints. Although the dry friction in each of these joints can be individually neglected, the integrated effect of twelve such elements becomes noticeable. Second, dynamical measurements might not be very accurate because the natural resonance frequency of the used structures is quite low even for relatively small dimensions. This dynamical response can be obviously modified with a proper mechanical design, but this increases the complexity of the sensor. This paper discusses the design and implementation of a touch pad based on a 6-axis force sensor and shows how the above limitations degrade its behavior. Moreover, it is shown how using a tensegrity structure both problems could be alleviated because ball-andsocket joints can be substituted by point contacts and the resonance frequency of the structure can be controlled by adjusting the static tensions of the tendons.Index Terms-touch pad, force sensor, wrench sensor.

No abstract

No abstract