1,721,006 research outputs found
A safe learning model reference adaptive controller for uncertain aircrafts models
No full textIn this paper, it is presented an approach for the design of a feedback Model Reference Adaptive Controller (MRAC) for uncertain linear systems with guaranteed evolution of the closed-loop error trajectories within a computable robust invariant set. The estimation of the UUB region has been derived exploiting robust quadratic stability arguments assuming bounded uncertainty on the state and input matrix as long as amplitude constrained adaptive control. In the scheme, a linear output feedback controller is augmented with an amplitude bounded adaptive control to improve the performance in the presence of significant modeling uncertainties. The robust invariant set design was performed solving a constrained convex optimization problem. The method allows the analysis of the joint effect of the modeling uncertainties and the adaptive control amplitude on the size of the UUB region, thus allowing the design of a safe MRAC scheme. An additional benefit is that no specific parameter adaptation algorithm is required, as long as the adaptive control output is confined within the predefined limits. For enforcing this confinement, the mechanism of adaptive control redistribution is introduced. A detailed simulation study was performed using the short period longitudinal dynamics of an F16 model to show the design steps and to highlight the benefits of the methodology
GOLN: Graph Object-based Localization Network
No full textIn the last decades, robotic localization has been mainly addressed with Visual Odometry (VO) or Simultaneous Localization and Mapping (SLAM) approaches, which usually provide an accurate metric precision. Despite the impressive results, these approaches have some shortcomings such as the amount of memory they require and the lack of robustness in non-ideal environments. Inspired by the human capabilities, in this paper we present a novel framework, named Graph Object-based Localization Network (GOLN), to address the topological localization problem with a novel approach, characterized by low memory requirements and robustness with respect to appearance. GOLN is based on a topological map, i.e., a graph, which is fed to a Graph Network (GN) along with global visual features of the environment and returns the estimation of the position node where the robot is located. Experiments have been performed in Unreal Engine (UE4) environments with a simulated ground robot, equipped with a monocular camera
A comprehensive case study of data-driven methods for robust aircraft sensor fault isolation
No full textRecent catastrophic events in aviation have shown that current fault diagnosis schemes may not be enough to ensure a reliable and prompt sensor fault diagnosis. This paper describes a comparative analysis of consolidated data-driven sensor Fault Isolation (FI) and Fault Estimation (FE) techniques using flight data. Linear regression models, identified from data, are derived to build primary and transformed residuals. These residuals are then implemented to develop fault isolation schemes for 14 sensors of a semi-autonomous aircraft. Specifically, directional Mahalanobis distance-based and fault reconstruction-based techniques are compared in terms of their FI and FE performance. Then, a bank of Bayesian filters is proposed to compute, in flight, the fault belief for each sensor. Both the training and the validation of the schemes are performed using data from multiple flights. Artificial faults are injected into the fault-free sensor measurements to reproduce the occurrence of failures. A detailed evaluation of the techniques in terms of FI and FE performance is presented for failures on the air-data sensors, with special emphasis on the True Air Speed (TAS), Angle of Attack (AoA), and Angle of Sideslip (AoS) sensors
Quantification of tolerable parametric and dynamic uncertainty for robust mrac systems
No full textAdaptive controllers have been proposed in a number of applications characterized by large modelling uncertainties in the open loop system. Despite this attractive feature it is not always immediate to guarantee, by design, a predictable closed-loop transient. Indeed the characterization of the transient response of a closed loop adaptive control system is still a challenging problem from the verification and validation standpoint. This issue is particularly relevant in the presence of unmodelled dynamics, time delays, disturbances and unmatched uncertainties. In this paper it is proposed a practical mythology for the quantification of the tolerable parametric and dynamic uncertainty in systems controlled by a Model Reference Adaptive Controller (MRAC) in the case the parameters that characterize the parametric and the dynamic uncertain as long as the controller adaptation weights are assumed unknown but bounded within a defined box domain. Given these bounded uncertainties, the MRAC design is formalized as a H2 controller design to guarantee minimum H2 gain between the reference model states and the tracking error states. Exploiting a quadratic Lyapunov function a parameter dependent LMI condition is derived whose feasibility guarantees a specified input-output H2 gain. This LMI condition can be exploited to quantify the tolerable matched and dynamic uncertainty and to evaluate the corresponding input-output H2 gain. The proposed approach has been applied to the design of a MRAC controller and to evaluate the tolerable uncertainties of a benchmark cart-pole system
Experimental Analysis of Neural Approaches for Synthetic Angle-of-Attack Estimation
No full textSynthetic sensors enable flight data estimation without devoted physical sensors. Within modern digital avionics, synthetic sensors can be implemented and used for several purposes such as analytical redundancy or monitoring functions. The angle of attack, measured at air data system level, can be estimated using synthetic sensors exploiting several solutions, e.g., model-based, data-driven, and model-free state observers. In the class of data-driven observers, multilayer perceptron neural networks are widely used to approximate the input-output mapping angle-of-attack function. Dealing with experimental flight test data, the multilayer perceptron can provide reliable estimation even though some issues can arise from noisy, sparse, and unbalanced training domain. An alternative is offered by regularization networks, such as radial basis function, to cope with training domain based on real flight data. The present work's objective is to evaluate performances of a single-layer feed-forward generalized radial basis function network for AoA estimation trained with a sequential algorithm. The proposed analysis is performed comparing results obtained using a multilayer perceptron network adopting the same training and validation data
Tire-road friction estimation and uncertainty assessment to improve electric aircraft braking system
No full textThe accurate online estimation of the road-friction coefficient is an essential feature for any advanced brake control system. In this study, a data-driven scheme based on a MLP Neural Net is proposed to estimate the optimum friction coefficient as a function of windowed slip-friction measurements. A stochastic NN weights drop-out mechanism is used to online estimate the confidence interval of the estimated best friction coefficient thus providing a characterization of the epistemic uncertainty associated to the NN block. Open loop and closed loop simulations of the landing phase of an aircraft on an unknown surface are used to show the potentiality and efficacy of the proposed robust friction estimation approach
A Data-Driven Slip Estimation Approach for Effective Braking Control under Varying Road Conditions
No full textThe performances of braking control systems for robotic platforms, e.g., assisted and autonomous vehicles, airplanes and drones, are deeply influenced by the road-tire friction experienced during the maneuver. Therefore, the availability of accurate estimation algorithms is of major importance in the development of advanced control schemes. The focus of this paper is on the estimation problem. In particular, a novel estimation algorithm is proposed, based on a multi-layer neural network. The training is based on a synthetic data set, derived from a widely used friction model. The open loop performances of the proposed algorithm are evaluated in a number of simulated scenarios. Moreover, different control schemes are used to test the closed loop scenario, where the estimated optimal slip is used as the set-point. The experimental results and the comparison with a model based baseline show that the proposed approach can provide an effective best slip estimation
Aircraft robust data-driven multiple sensor fault diagnosis based on optimality criteria
No full textA general robust data-driven scheme for the Fault Detection, Isolation and Estimation of multiple sensor faults is proposed and validated using multi-flight data records. Robustness to modelling uncertainty and noise is achieved through an optimized data-driven design of the three blocks that constitute the scheme. First, a robust Fault Detection (FD) filter given by the linear combination of previously identified Analytical Redundancy Relationships (AARs) is derived as the solution of a multi-objective optimization where the minimum fault sensitivity is maximized while the standard deviation (STD) of the filtered error, in nominal condition, is minimized. Then, a Fault Pre-Isolation (FPI) block is introduced to select a restricted number of sensors containing with high likelihood the subset of the faulty sensors. In this phase, robustness is achieved through the data-driven design of a redundant number of Multiple-ARRs and a voting logic. Finally, the robust Fault Isolation (FI) is achieved relying on the design of a large collection of additional AARs whose fault signatures are specifically designed to optimize, at the same time, noise immunity while maximizing the decoupling of the (pre-isolated) fault directions. A procedure based on fault amplitude reconstruction is proposed to isolate the set of faulty sensors sequentially. The proposed scheme has been applied to the design of a multiple Fault Diagnosis scheme for a set of 8 sensors of a semi-autonomous aircraft basing on multi-flight data. Validation results are compared with state-of-the-art multiple Fault Diagnosis schemes
Data-driven schemes for robust fault detection of air data system sensors
No full textFailures of the air data system due to exceptional weather conditions have shown to be among the leading causes of aviation accidents. In this respect, most of aircraft Pitot tube failure detection schemes rely on mathematical models and simplified assumptions on the uncertain parameters and disturbances. These methods typically require 'ad hoc' time-consuming tuning procedures that may produce unreliable performance when validated with actual flight data. In this paper, a complete semiautomated data-driven approach is introduced to select the model regressors, to identify NARX input-output prediction models, to set up robust fault detection filters and to compute fault detection thresholds. To cope with time-dependent and flight-dependent levels of uncertainties online model adaption mechanisms are introduced to limit the critical problem of minimizing the false alarm rate. Extensive validation tests have been conducted using actual flight data of a P92 Tecnam aircraft through the introduction of artificially injected hard and soft failure of the Pitot tube sensor. The approach showed to be remarkably robust in terms of false alarms while maintaining fault detectability to faults of amplitudes less than 1 m/s
- …
