arXiv subjects

Didier Henrion

Publications and source records attributed to Didier Henrion.

At least 109 records · Page 6Linked to original sources

arXiv2014

Uniform sample generation in semialgebraic sets

We propose efficient techniques for generating independent identically distributed uniform random samples inside semialgebraic sets. The proposed algorithm leverages recent results on the approximation of indicator functions by polynomials %\cite{DabHen:13} to develop acceptance/rejection based sample generation algorithms with guaranteed performance in terms of rejection rate (the number of samples that should be generated in order to obtain an accepted sample). Moreover, the {acceptance} rate is shown to be is asymptotically optimal, in the sense that it tends to one (all samples accepted) as the degree of the polynomial approximation increases. The performance of the proposed method is illustrated by a numerical example.

math.OC↗

arXiv2014

Hand-Eye and Robot-World Calibration by Global Polynomial Optimization

The need to relate measurements made by a camera to a different known coordinate system arises in many engineering applications. Historically, it appeared for the first time in the connection with cameras mounted on robotic systems. This problem is commonly known as hand-eye calibration. In this paper, we present several formulations of hand-eye calibration that lead to multivariate polynomial optimization problems. We show that the method of convex linear matrix inequality (LMI) relaxations can be used to effectively solve these problems and to obtain globally optimal solutions. Further, we show that the same approach can be used for the simultaneous hand-eye and robot-world calibration. Finally, we validate the proposed solutions using both synthetic and real datasets.

cs.CV↗

arXiv2013

Convex computation of the region of attraction of polynomial control systems

We address the long-standing problem of computing the region of attraction (ROA) of a target set (e.g., a neighborhood of an equilibrium point) of a controlled nonlinear system with polynomial dynamics and semialgebraic state and input constraints. We show that the ROA can be computed by solving an infinite-dimensional convex linear programming (LP) problem over the space of measures. In turn, this problem can be solved approximately via a classical converging hierarchy of convex finite-dimensional linear matrix inequalities (LMIs). Our approach is genuinely primal in the sense that convexity of the problem of computing the ROA is an outcome of optimizing directly over system trajectories. The dual infinite-dimensional LP on nonnegative continuous functions (approximated by polynomial sum-of-squares) allows us to generate a hierarchy of semialgebraic outer approximations of the ROA at the price of solving a sequence of LMI problems with asymptotically vanishing conservatism. This sharply contrasts with the existing literature which follows an exclusively dual Lyapunov approach yielding either nonconvex bilinear matrix inequalities or conservative LMI conditions. The approach is simple and readily applicable as the outer approximations are the outcome of a single semidefinite program with no additional data required besides the problem description.

math.OC↗

arXiv2013

Probabilistic and Set-based Model Invalidation and Estimation using LMIs

Probabilistic and set-based methods are two approaches for model invalidation, parameter and state estimation. Both classes of methods use different types of data, i.e. deterministic or probabilistic data, which allow different statements and applications. Ideally, however, all available data should be used in estimation and model invalidation methods. This paper presents an estimation and model invalidation framework combining set-based and probabilistically uncertain data for polynomial continuous-time systems. In particular, uncertain data on the moments and the support is used without the need to make explicit assumptions on the type of probability densities. The paper derives pointwise-in-time outer approximations of the moments of the probability densities associated with the states and parameters of the system. These approximations can be interpreted as guaranteed confidence intervals for the moment estimates. Furthermore, guaranteed bounds on the probability masses on subsets are derived and allow an estimation of the unknown probability densities. To calculate the estimates, the dynamics of the probability densities of the state trajectories are found by occupation measures of the nonlinear dynamics. This allows the construction of an infinite-dimensional linear program which incorporates the set- and moment-based data. This linear program is relaxed by a hierarchy of LMI problems providing, as shown elsewhere, an almost uniformly convergent sequence of outer approximations of the estimated sets. The approach is demonstrated with numerical examples.

math.OC↗

arXiv2013

Optimization on linear matrix inequalities for polynomial systems control

Many problems of systems control theory boil down to solving polynomial equations, polynomial inequalities or polyomial differential equations. Recent advances in convex optimization and real algebraic geometry can be combined to generate approximate solutions in floating point arithmetic. In the first part of the course we describe semidefinite programming (SDP) as an extension of linear programming (LP) to the cone of positive semidefinite matrices. We investigate the geometry of spectrahedra, convex sets defined by linear matrix inequalities (LMIs) or affine sections of the SDP cone. We also introduce spectrahedral shadows, or lifted LMIs, obtained by projecting affine sections of the SDP cones. Then we review existing numerical algorithms for solving SDP problems. In the second part of the course we describe several recent applications of SDP. First, we explain how to solve polynomial optimization problems, where a real multivariate polynomial must be optimized over a (possibly nonconvex) basic semialgebraic set. Second, we extend these techniques to ordinary differential equations (ODEs) with polynomial dynamics, and the problem of trajectory optimization (analysis of stability or performance of solutions of ODEs). Third, we conclude this part with applications to optimal control (design of a trajectory optimal w.r.t. a given functional). For some of these decision and optimization problems, it is hoped that the numerical solutions computed by SDP can be refined a posteriori and certified rigorously with appropriate techniques.

math.OC↗

arXiv2013

Design of Marx generators as a structured eigenvalue assignment

We consider the design problem for a Marx generator electrical network, a pulsed power generator. The engineering specification of the design is that a suitable resonance condition is satisfied by the circuit so that the energy initially stored in a number of storage capacitors is transferred in finite time to a single load capacitor which can then store the total energy and deliver the pulse. We show that the components design can be conveniently cast as a structured real eigenvalue assignment with significantly lower dimension than the state size of the Marx circuit. Then we comment on the nontrivial nature of this structured real eigenvalue assignment problem and present two possible approaches to determine its solutions. A first symbolic approach consists in the use of Gröbner basis representations, which allows us to compute all the (finitely many) solutions. A second approach is based on convexification of a nonconvex optimization problem with polynomial constraints. We show that the symbolic method easily provides solutions for networks up to six stages while the numerical method can reach up to seven and eight stages. We also comment on the conjecture that for any number of stages the problem has finitely many solutions, which is a necessary assumption for the proposed methods to converge. We regard the proof of this conjecture as an interesting challenge of general interest in the real algebraic geometry field.

math.OC↗

arXiv2013

Convex computation of the maximum controlled invariant set for polynomial control systems

We characterize the maximum controlled invariant (MCI) set for discrete- as well as continuous-time nonlinear dynamical systems as the solution of an infinite-dimensional linear programming problem. For systems with polynomial dynamics and compact semialgebraic state and control constraints, we describe a hierarchy of finite-dimensional linear matrix inequality (LMI) relaxations whose optimal values converge to the volume of the MCI set; dual to these LMI relaxations are sum-of-squares (SOS) problems providing a converging sequence of outer approximations to the MCI set. The approach is simple and readily applicable in the sense that the approximations are the outcome of a single semidefinite program with no additional input apart from the problem description. A number of numerical examples illustrate the approach.

math.OC↗

arXiv2013

Estimation of consistent parameter sets for continuous-time nonlinear systems using occupation measures and LMI relaxations

Obtaining initial conditions and parameterizations leading to a model consistent with available measurements or safety specifications is important for many applications. Examples include model (in-)validation, prediction, fault diagnosis, and controller design. We present an approach to determine inner- and outer-approximations of the set containing all consistent initial conditions/parameterizations for nonlinear continuous-time systems. These approximations are found by occupation measures that encode the system dynamics and measurements, and give rise to an infinite-dimensional linear program. We exploit the flexibility and linearity of the decision problem to incorporate uncertain-but-bounded and pointwise-in-time state and output constraints, a feature which was not addressed in previous works. The infinite-dimensional linear program is relaxed by a hierarchy of LMI problems that provide certificates in case no consistent initial condition/parameterization exists. Furthermore, the applied LMI relaxation guarantees that the approximations converge (almost uniformly) to the true consistent set. We illustrate the approach with a biochemical reaction network involving unknown initial conditions and parameters.

math.OC↗

arXiv2013

Optimal switching control design for polynomial systems: an LMI approach

We propose a new LMI approach to the design of optimal switching sequences for polynomial dynamical systems with state constraints. We formulate the switching design problem as an optimal control problem which is then relaxed to a linear programming (LP) problem in the space of occupation measures. This infinite-dimensional LP can be solved numerically and approximately with a hierarchy of convex finite-dimensional LMIs. In contrast with most of the existing work on LMI methods, we have a guarantee of global optimality, in the sense that we obtain an asympotically converging (i.e. with vanishing conservatism) hierarchy of lower bounds on the achievable performance. We also explain how to construct an almost optimal switching sequence.

math.OC↗

arXiv2013

Moment LMI approach to LTV impulsive control

In the 1960s, a moment approach to linear time varying (LTV) minimal norm impulsive optimal control was developed, as an alternative to direct approaches (based on discretization of the equations of motion and linear programming) or indirect approaches (based on Pontryagin's maximum principle). This paper revisits these classical results in the light of recent advances in convex optimization, in particular the use of measures jointly with hierarchy of linear matrix inequality (LMI) relaxations. Linearity of the dynamics allows us to integrate system trajectories and to come up with a simplified LMI hierarchy where the only unknowns are moments of a vector of control measures of time. In particular, occupation measures of state and control variables do not appear in this formulation. This is in stark contrast with LMI relaxations arising usually in polynomial optimal control, where size grows quickly as a function of the relaxation order. Jointly with the use of Chebyshev polynomials (as a numerically more stable polynomial basis), this allows LMI relaxations of high order (up to a few hundreds) to be solved numerically.

math.OC↗

arXiv2012

Measures and LMIs for optimal control of piecewise-affine systems

This paper considers the class of deterministic continuous-time optimal control problems (OCPs) with piecewise-affine (PWA) vector field, polynomial Lagrangian and semialgebraic input and state constraints. The OCP is first relaxed as an infinite-dimensional linear program (LP) over a space of occupation measures. This LP, a particular instance of the generalized moment problem, is then approached by an asymptotically converging hierarchy of linear matrix inequality (LMI) relaxations. The relaxed dual of the original LP returns a polynomial approximation of the value function that solves the Hamilton-Jacobi-Bellman (HJB) equation of the OCP. Based on this polynomial approximation, a suboptimal policy is developed to construct a state feedback in a sample-and-hold manner. The results show that the suboptimal policy succeeds in providing a stabilizing suboptimal state feedback law that drives the system relatively close to the optimal trajectories and respects the given constraints.

math.OC↗

arXiv2012

Minimum volume semialgebraic sets for robust estimation

Motivated by problems of uncertainty propagation and robust estimation we are interested in computing a polynomial sublevel set of fixed degree and minimum volume that contains a given semialgebraic set K. At this level of generality this problem is not tractable, even though it becomes convex e.g. when restricted to nonnegative homogeneous polynomials. Our contribution is to describe and justify a tractable L^1-norm or trace heuristic for this problem, relying upon hierarchies of linear matrix inequality (LMI) relaxations when K is semialgebraic, and simplifying to linear programming (LP) when K is a collection of samples, a discrete union of points.

math.OC↗

arXiv2012

Inner approximations of the region of attraction for polynomial dynamical systems

In a previous work we developed a convex infinite dimensional linear programming (LP) approach to approximating the region of attraction (ROA) of polynomial dynamical systems subject to compact basic semialgebraic state constraints. Finite dimensional relaxations to the infinite-dimensional LP lead to a truncated moment problem in the primal and a polynomial sum-of-squares problem in the dual. This primal-dual linear matrix inequality (LMI) problem can be solved numerically with standard semidefinite programming solvers, producing a hierarchy of outer (i.e. exterior) approximations of the ROA by polynomial sublevel sets, with a guarantee of almost uniform and set-wise convergence. In this companion paper, we show that our approach is flexible enough to be modified so as to generate a hierarchy of polynomial inner (i.e.\,interior) approximations of the ROA with similar convergence guarantees.

math.OC↗

arXiv2012

Mean squared error minimization for inverse moment problems

We consider the problem of approximating the unknown density $u\in L^2(Ω,λ)$ of a measure $μ$ on $Ω\subset\R^n$, absolutely continuous with respect to some given reference measure $λ$, from the only knowledge of finitely many moments of $μ$. Given $d\in\N$ and moments of order $d$, we provide a polynomial $p_d$ which minimizes the mean square error $\int (u-p)^2dλ$ over all polynomials $p$ of degree at most $d$. If there is no additional requirement, $p_d$ is obtained as solution of a linear system. In addition, if $p_d$ is expressed in the basis of polynomials that are orthonormal with respect to $λ$, its vector of coefficients is just the vector of given moments and no computation is needed. Moreover $p_d\to u$ in $L^2(Ω,λ)$ as $d\to\infty$. In general nonnegativity of $p_d$ is not guaranteed even though $u$ is nonnegative. However, with this additional nonnegativity requirement one obtains analogous results but computing $p_d\geq0$ that minimizes $\int (u-p)^2dλ$ now requires solving an appropriate semidefinite program. We have tested the approach on some applications arising from the reconstruction of geometrical objects and the approximation of solutions of nonlinear differential equations. In all cases our results are significantly better than those obtained with the maximum entropy technique for estimating $u$.

math.OC↗

arXiv2012

Measures and LMI for space launcher robust control validation

We describe a new temporal verification framework for safety and robustness analysis of nonlinear control laws, our target application being a space launcher vehicle. Robustness analysis, formulated as a nonconvex nonlinear optimization problem on admissible trajectories corresponding to piecewise polynomial dynamics, is relaxed into a convex linear programming problem on measures. This infinite-dimensional problem is then formulated as a generalized moment problem, which allows for a numerical solution via a hierarchy of linear matrix inequality relaxations solved by semidefinite programming. The approach is illustrated on space launcher vehicle benchmark problems, in the presence of closed-loop nonlinearities (saturations and dead-zones) and axis coupling.

math.OC↗

arXiv2012

Inner approximations for polynomial matrix inequalities and robust stability regions

Following a polynomial approach, many robust fixed-order controller design problems can be formulated as optimization problems whose set of feasible solutions is modelled by parametrized polynomial matrix inequalities (PMI). These feasibility sets are typically nonconvex. Given a parametrized PMI set, we provide a hierarchy of linear matrix inequality (LMI) problems whose optimal solutions generate inner approximations modelled by a single polynomial sublevel set. Those inner approximations converge in a strong analytic sense to the nonconvex original feasible set, with asymptotically vanishing conservatism. One may also impose the hierarchy of inner approximations to be nested or convex. In the latter case they do not converge any more to the feasible set, but they can be used in a convex optimization framework at the price of some conservatism. Finally, we show that the specific geometry of nonconvex polynomial stability regions can be exploited to improve convergence of the hierarchy of inner approximations.

math.OC↗

arXiv2012

Convex inner approximations of nonconvex semialgebraic sets applied to fixed-order controller design

We describe an elementary algorithm to build convex inner approximations of nonconvex sets. Both input and output sets are basic semialgebraic sets given as lists of defining multivariate polynomials. Even though no optimality guarantees can be given (e.g. in terms of volume maximization for bounded sets), the algorithm is designed to preserve convex boundaries as much as possible, while removing regions with concave boundaries. In particular, the algorithm leaves invariant a given convex set. The algorithm is based on Gloptipoly 3, a public-domain Matlab package solving nonconvex polynomial optimization problems with the help of convex semidefinite programming (optimization over linear matrix inequalities, or LMIs). We illustrate how the algorithm can be used to design fixed-order controllers for linear systems, following a polynomial approach.

math.OC↗

arXiv2011

Linear Control of Time-Domain Constrained Systems

This paper presents a general framework for the design of linear controllers for linear systems subject to time-domain constraints. The design framework exploits sums-of-squares techniques to incorporate the time-domain constraints on closed-loop signals and leads to conditions in terms of linear matrix inequalities (LMIs). This control design framework offers, in addition to constraint satisfaction, also the possibility of including an optimization objective that can be used to minimize steady state (tracking) errors, to decrease the settling time, to reduce overshoot and so on. The effectiveness of the framework is shown via a numerical example.

math.OC↗