Marko Tanaskovic

Adaptive receding horizon control for constrained MIMO systems

An adaptive control algorithm for open-loop stable, constrained, linear, multiple input multiple output systems is presented. The proposed approach can deal with both input and output constraints, as well as measurement noise and output disturbances. The adaptive controller consists of an iterative set membership identification algorithm, that provides a set of candidate plant models at each time step, and a model predictive controller, that enforces input and output constraints for all the plants inside the model set. The algorithm relies only on the solution of standard convex optimization problems that are guaranteed to be recursively feasible. The experimental results obtained by applying the proposed controller to a quad-tank testbed are presented.

Wrong turn - No dead end: A stochastic pedestrian motion model

This paper addresses the use of social behavior models for the prediction of a pedestrians future motion. Recently, such models have been shown to outperform simple constant velocity models in cases where data association becomes ambiguous, e.g. in case of occlusion, bad image quality, or low frame rates. However, to account for the multiple alternatives a pedestrian can choose from, one has to go beyond the currently available deterministic models. To this end, we propose a stochastic extension of a recently proposed simulation-based motion model. This new instantiation can cater for the possible behaviors in an entire scene in a multi-hypothesis approach, using a principled modeling of uncertainties. In a set of experiments for prediction and template-based tracking, we compare it to a deterministic instantiation and investigate the general value of using an advanced motion prior in tracking.

Data-driven control of nonlinear systems: An on-line direct approach☆

Abstract A data-driven method to design reference tracking controllers for nonlinear systems is presented. The technique does not derive explicitly a model of the system, rather it delivers directly a time-varying state-feedback controller by combining an on-line and an off-line scheme. Like in other on-line algorithms, the measurements collected in closed-loop operation are exploited to modify the controller in order to improve the tracking performance over time. At the same time, a predictable closed-loop behavior is guaranteed by making use of a batch of available data, which is a feature of off-line algorithms. The feedback controller is parameterized with kernel functions and the design approach exploits results in set membership identification and learning by projections. Under the assumptions of Lipschitz continuity and stabilizability of the system’s dynamics, it is shown that if the initial batch of data is informative enough, then the resulting closed-loop system is guaranteed to be finite gain stable. In addition to the main theoretical properties of the approach, the design algorithm is demonstrated experimentally on a water tank system.

On the optimal worst-case experiment design for constrained linear systems

The problem of experiment design for constrained linear systems with multiple inputs is addressed. A parametric model of the system is considered. The presented theoretical results provide a guideline on how to design experiments that minimize the worst-case identification error, as measured by the radius of information of the set of feasible model parameters, calculated in any norm. In addition, it is shown that an alternative, simpler approach can be employed when input constraints are symmetric and the worst-case identification error is minimized in either 1 - or ∞ -norm. For such cases, on the basis of the derived results, a computationally tractable algorithm for the experiment design is proposed. The presented results are valid for a general model representation, which admits the commonly used finite impulse response model as a special case. The features of the presented method are illustrated in a numerical example.

Experimental testing of an adaptive model predictive controller on a quad-tank system

A recently proposed adaptive model predictive control algorithm is implemented in real time and its performance is verified experimentally on a quad-tank testbed. The algorithm relies on an iterative set membership identification procedure in order to provide a set of possible plant models at each time step. This set is then exploited by the model predictive controller in order to robustly enforce output constraints. The experimental results show that the algorithm provides good reference tracking performance while robustly satisfying output constraints for different operating conditions.

Adaptive Model Predictive Control for Constrained MIMO Systems

Abstract An adaptive output feedback control algorithm for constrained multiple input multiple output linear systems is proposed, able to cope with input and output constraints, output disturbances and measurement noise. The approach relies on a real-time set membership identification algorithm to provide bounds on the predicted plant outputs. These bounds are exploited in a receding horizon control strategy that guarantees recursive satisfaction of constraints. The algorithm yields offset free reference tracking in case of constant output disturbance and zero measurement noise. The effectiveness of the proposed approach is illustrated on a numerical example.

Worst-case experiment design for constrained MISO systems

The problem of optimal worst-case experiment design for constrained linear systems with multiple inputs represented by a parametric model is addressed. A theoretical result is derived, which provides an insight on how to design experiments that minimize the worst-case identification error in ∞- and 1-norm when the input constraints are symmetric. The presented result is valid for a general model parametrization that admits the commonly used finite impulse response model as a special case. Based on this result a computationally tractable algorithm for the worst-case experiment design is proposed. Its advantages over a more standard experiment design approach are illustrated in a numerical example.