Fabien Claveau

Modeling the Visual and Motor Control of Steering With an Eye to Shared-Control Automation

This paper describes a newly developed driver model that focuses on the control of car steering. The model represents visual anticipation of road curvature and compensation of lateral positioning error. It also incorporates a neuromuscular system, inspired by Hoult and Cole (2008), including an internal model of the steering system compliance, muscle co-activation by α and γ signals and the stretch reflex. Preliminary driving simulator experiments with five participants showed that the identification of model parameters yielded consistent results. Moreover, the model was able to steer the driving simulator by itself and showed a behavior similar to that of the human driver who provided the data for parameter identification. This model may be used for the design of automation for shared control of steering.

Direct and Steering Tilt Robust Control of Narrow Vehicles

Narrow tilting vehicles (NTVs) are the convergence of a car and a motorcycle. They are expected to be the new generation of city cars considering their practical dimensions and lower energy consumption. However, due to their height-to-breadth ratio, in order to maintain lateral stability, NTVs should tilt when cornering. Unlike the motorcycle, where the driver tilts the vehicle himself, the tilting of an NTV should be automatic. Two tilting systems are available, i.e., direct tilt control (DTC) and steering tilt control (STC), the combined action of these two systems being certainly the key to considerably improve NTV dynamic performances. In this paper, multivariable control tools (H2 methodology) are used to design, in a systematic way, lateral assistance controllers driving DTC, STC, or both DTC/STC systems. A three-degree-of-freedom (DoF) model of the vehicle is used, as well as a model of the steering signal, leading to a two-DoF low-order controller with an efficient feedforward anticipative part. Taking advantage of all the available measurements on NTVs, the lateral acceleration is directly regulated. Finally, a gain-scheduling solution is provided to make the DTC, STC, and DTC/STC controllers robust to longitudinal speed variations.

A Lateral Control Strategy for Narrow Tilting Commuter Vehicle Based on the Perceived Lateral Acceleration.

Abstract Narrow tilting commuter vehicles are expected to be the new generation of city cars, considering their practical dimensions and lower energy consumption. But their dimensions increase their tendency to overturn during cornering, facing lateral acceleration. This problem can be solved by tilting the vehicle in a way that reduces the perceived lateral acceleration at the cabin during cornering, as it is seen on two wheeled vehicles. So far in literature, the corresponding tilting angle is computed, and the control strategy aims at reaching this desired angle. This paper, presents another control approach, achieving a direct control of the perceived acceleration. The proposed strategy is interesting since it is simple to implement, takes advantage of accelerometers and gyroscopes measures, and valid on roads with non zero banking angle. As the lateral speed is not practically measured, the paper proposes, for comparison purpose, different (original) ways to get its value without deteriorating robustness. Finally, the proposed solution is a simple robust state feedback controller exploiting all the available measures.

Design and optimization of an hybrid sailboat by a power modeling approach

In this paper an original modeling approach is proposed. This approach is based on power exchanges and highlights the most influential parameters of a complex system, without having to choose particular technological solutions. This is thus a particularly well suited solution for a first and global design or for the validation of a concept. This approach has been efficiently used for the optimal sizing and control of a hybrid sailboat which comprises a thermal engine and two electric motors in serie configuration associated with sails. The control strategy is organized in two levels: a global one, which defines power objectives, and a local one, which maximizes the efficiency, while guaranteeing the security of each component.

Design of a two DOF gain scheduled frequency shaped LQ controller for Narrow Tilting Vehicles

Narrow Tilting Vehicles (NTVs) are the convergence of a car and a motorcycle. They are expected to be the new generation of city cars considering their practical dimensions and lower energy consumption. But considering their height to breadth ratio, in order to maintain lateral stability, NTVs should tilt when cornering. Unlike the motorcycles case, where the driver tilts the vehicle himself, the tilting of an NTV should be automatic. The control objective in this paper is to reduce the perceived lateral acceleration through tilting, what would increase lateral stability and also the driver comfort with regard to centrifugal acceleration. A two degree of freedom tilting controller is proposed in the paper, using all the measurement available, controlling directly the perceived lateral acceleration (measured), while considering the steering angle as the main disturbance source. The control problem is recast as an optimal H2 problem, considering a model of the typical steering angle in addition to the vehicle model. The steering angle model allows predicting the near future of the vehicle trajectory, leading to an improved result in terms of energy requirements. Although initially designed for a given constant speed, the LPV controller is finally proposed with guarantees of good performance for a large range of speed and acceleration.

AN EXTENDED STATE-FEEDBACK H2 CONTROLLER FOR DESCRIPTOR SYSTEMS

Abstract This paper investigates the design of an “extended” state-feedback H 2 controller for continuous-time descriptor systems. In such problem, the controller, which is in descriptor form, is found such that : - The closed-loop transfer matrix is strictly proper and has minimum H 2 -norm. - Only the internal stability of the physical system (and not the overall descriptor standard model) is required. An explicit characterization of the solution based on a generalized algebraic Riccati equation and two Sylvester equations is given. Under some additional assumptions, a LMI-based solution is proposed. A numerical example is also included to show the applicability of the proposed results.

Energy Management in multi-consumers multi-sources System: A Practical Framework

Abstract The paper develops a methodology to design a practical Multi-sources and Multi-consumers energy management systems (EMS), applicable in particular to transport systems. A global EMS optimization problem is formulated underlining generics criterion and constraints. Remarking that a modularity property is essential to add easily (i.e. without redesigning the whole problem) new energy consumers (or sources), a new EMS structure is presented, splitting the problem into two independent sub-problems. In compensation of sub-optimality, the computing burden is lighten and robustness to energy shortage enhanced. The framework is introduced through the EMS design of an hybrid vehicle transporting conditioned merchandises.

OBSERVER-BASED GAIN SCHEDULING H∞ CONTROLLER FOR A 3 MOTORS WEB TRANSPORT SYSTEM

Abstract This paper deals with the robust gain scheduled control problem of a web transport system with winder and unwinder for elastic material. Precisely, the objective consists to control accurately the tension along the web and its velocity in order to prevent the occurrence of web break or fold. Due to the wide-range variation of the radius and inertia of the rollers the system dynamics change considerably during the winding/unwinding process. A Linear Parameter Varying (LPV) control strategy is derived from the interpolation of the observer state- feedback gains of H ∞ robust controllers. The resulting controllers are based on unbiased LPV observers and perform well during the whole winding/unwinding process.

A two-layer LPV based control strategy for input and state constrained problem: Application to energy management

This paper proposes a pragmatic solution to solve input and state constrained control problems. Taking benefits from a two-layer hierarchical architecture, in particular by working at a different period at each level, the general idea is to combine an explicit LPV controller ensuring the regulation task, and a predictive control at the upper level to comply with the active constraints. In practice, the two layers are connected as the external loop drives the varying parameter of the inner loop. The proposed scheme is well suited mainly with control problems whose constraints violation risks may be predicted sufficiently in advance. The energy management of a hybrid vehicle is finally considered to illustrate the applicability.

Non-linear control of a Narrow Tilting Vehicle.

Narrow Tilting Vehicles (NTVs) are the convergence of a car and a motorcycle. They are expected to be the new generation of city cars considering their practical dimensions and lower energy consumption. But considering their height to breadth ratio, in order to maintain lateral stability, NTVs should tilt when cornering. Unlike the motorcycles case, where the driver tilts the vehicle himself, the tilting of an NTV should be automatic. Two tilting systems are available; Direct and Steering Tilt Control, the combined action of these two systems being certainly the key to improve considerably NTVs dynamic performances. Focusing on the lateral dynamic of NTVs, multivariable control strategies based on linear robust control theory, were already proposed in the literature, assuming decoupling with the longitudinal dynamic. In this paper a 4 DoF model of the main longitudinal and lateral dynamics is considered, and its differential flatness is demonstrated. The three flat outputs have furthermore a particular physical meaning, making possible the design of a simple external control loop complying with the driver demands.