Jorge Estrela da Silva

A CONTROL ARCHITECTURE FOR MULTIPLE SUBMARINES IN COORDINATED SEARCH MISSIONS

A control architecture for executing multi-vehicle search algorithms is presented. The proposed hierarchical structure consists of three control layers: maneuver controllers, vehicle supervisors and team controllers. The system model is described as a dynamic network of hybrid automata in the programming language Shift and allows reasoning about specification and dynamical properties in a formal setting. The particular search problem that is studied is that of finding the minimum of a scalar field using a team of autonomous submarines. As an illustration, a coordination scheme based on the Nelder-Mead simplex optimization algorithm is presented and illustrated through simulations.

A reconfigurable mission control system for underwater vehicles

This paper describes the mission control software used in the LSTS/FEUP underwater vehicles. This software follows the guidelines of the generalized vehicle architecture, adapts the original idea to encompass the current application requirements and constitutes a first implementation. The work is focused on the design and implementation of an application that can be easily adapted to different vehicle configurations or even to different vehicles. One of the desired goals was to enhance software reusability and to establish a development environment that allows developers with a minimal knowledge of coding details to upgrade the application. To assist this purpose, a CASE tool, which provides modern software development techniques, was used. A simulation environment was also developed whose purpose is to test the applications and to detect possible malfunctions before they occur during mission execution.

A dynamic programming approach for the motion control of autonomous vehicles

The problem of 2-dimensional vehicle path following under adversarial behavior is considered. The objective is to keep the cross-track error to the reference path inside a given tolerance interval. The adversarial behavior models system uncertainty and unknown or poorly estimated bounded disturbances. The first step to that objective is the computation of an invariant set, namely the maximal set of states where the following property holds: if the vehicle starts inside the set there exists a control law such that the cross-track will never exceed the tolerance interval. This is done through dynamic programming. Two modes of operation are then considered for the control system: when the vehicle is inside the invariant set, the objective is to stay inside it while minimizing a combination of the actuation effort and cross-track error; otherwise, the objective becomes to reach the invariant set in minimum time. Each mode corresponds to a different optimal control problem which is dealt independently thus leading to a specific control law.

Hierarchical search strategy for a team of autonomous vehicles

Abstract A multi-vehicle search strategy based on the simplex algorithm is proposed. The strategy is mapped into a hierarchical scheme with three layers. The upper layer is described by a discrete-event system. The output of this layer is a set of way-points for the vehicles and it is used by the middle layer in order to coordinate the motion of the vehicles. The lower layer drives each vehicle to a way-point. The paper compares two possible discrete coordination strategies: one minimizes the total travelled distance and the other prevents trajectories to intersect. Minimization of vehicle intercommunication is also studied.

Dynamic programming based feedback control for systems with switching costs

This paper is concerned with the synthesis of optimal feedback controllers for nonlinear continuous-time systems with switching costs. The design approach is based on dynamic programming techniques. Robustness with respect to model uncertainty and bounded disturbances is encompassed by considering an adversarial input. The problem is solved in the framework of zero-sum differential games. Controller specifications include state and input constraints. A numerical algorithm for the synthesis of the controller is described.

Dynamic Programming Techniques for Feedback Control

Abstract We consider a dynamic programming approach for solving optimal control problems of sampled continuous-time systems. Robustness to bounded noise and model uncertainties is provided by formulating the problem in the framework of differential games. We use a semi-Lagrangian scheme for the computation of the value function and the state feedback control law for constrained infinite horizon optimal control problems. The approach is illustrated with the optimal control of linear systems and nonlinear systems subject to bounded disturbances.

A dynamic programming approach for the control of autonomous vehicles on planar motion

The problem of path following for autonomous vehicles under adversarial behavior is considered. The objective is to keep the cross-track error to the reference path inside a given tolerance interval. The adversarial behavior models system uncertainty and unknown or poorly estimated bounded disturbances. The first step to that objective is the computation of an invariant set, namely the maximal set of states that the vehicle may enter while ensuring that the cross-track error will never exceed the tolerance interval. This is done through dynamic programming. Two modes of operation are then considered: when the vehicle is inside the invariant set, the objective is to stay inside it while minimizing a combination of the actuation effort and cross-track error; otherwise, the objective becomes to reach the invariant set in minimum time. Each mode corresponds to a different optimal control problem which is dealt independently; thus, each mode has a corresponding control law. We discuss efficient ways of computing and implementing those control laws on currently available computational systems. For the purpose of the dynamic programming approach, the autonomous vehicles are modeled as an unicycle. Simulations with a six degree of freedom nonlinear model of an autonomous submarine are performed in order to illustrate the robustness of the control strategy.

Trajectory optimization for data retrieval applications: it is all about where/who loads the mule

Abstract This work analyses the problem of data collection in sensor networks with disconnected clusters of sensors leveraging autonomous vehicles and intra-cluster sensor cooperation. The later allows for the vehicles trajectory to be planned in terms of visiting each cluster instead of visiting individual sensors with important repercussions on the computational complexity of the problem. This paper proposes a dynamic programming methodology to solve the data collection problem in optimal time assuming that the cost of sensor cooperation is negligible. Our approach is novel in that it considers (i) intra-cluster cooperation, allowing communication to a subset of the nodes in each cluster, (ii) a communication rate to each cluster depends on the location of the vehicle, i.e., the data acquisition rate is state dependent, and (iii) that the amount of information to be collected on each sensor cluster may be either known or unknown in advance. Although the problem is cast in the context of data collection in a set of isolated sensor networks, the developed methodology may be used in other applications of resource acquisition or consumption. Simulation results are provided for data collection scenarios with 3 sensor clusters.

Experimental results with value function based control of an AUV

Abstract This paper reports experimental results obtained with a value function based path following controller for Autonomous Underwater Vehicles (AUV). The value functions are computed using numerical methods for the dynamic programming (DP) equations. We consider the case of planar operation of an AUV with actuation surfaces (fins). A reduced nonlinear model with 4 state variables is considered: cross-track error, angle with respect to path, angular velocity and fins position. The objective of our control approach is twofold: to ensure given hard constraints on the maximum cross-track error in the presence of bounded disturbances; to provide a balance between the controller performance and the rate of change in the fins position, in order to avoid premature wear of the servomotors. The underwater experiments are performed with an AUV from Porto University.

Reachability analysis of dynamic programming based controlled systems

This paper extends the algorithm proposed in [1]. In [1], the numerical approximation of the solution of dynamic programming problems is used as a candidate Lyapunov function VL(x), in order to prove convergence of the closed loop system to a given target. That is done in the framework of sampled data systems. The described algorithm provides only a sufficient condition, which in many cases is not verified due to the deviations introduced by the approximations. In this paper, the previous algorithm is extended in order to compute the decrease in the candidate Lyapunov over multiple sampling periods, i.e., VL(x(k + n)) - VL(x(k)), with n ϵ N. Thus, it is expected that the verification procedure becomes more robust with respect to the deviations introduced by the numerical approximation of the control law and candidate Lyapunov function. The core of the algorithm is based on linear programming with polytopic convex constraints.