Ashwin Carvalho

Automated Driving The Role of Forecasts and Uncertainty - A Control Perspective

Abstract Driving requires forecasts. Forecasted movements of objects in the driving scene are uncertain. Inevitably, decision and control algorithms for autonomous driving need to cope with such uncertain forecasts. In assisted driving, the uncertainty in the human/vehicle interaction further increases the complexity of the control design task. Our research over the past ten years has focused on control design methods which systematically handle uncertain forecasts for autonomous and semi-autonomous vehicles. This paper presents an overview of our findings and discusses relevant aspects of our recent results.

Linear model predictive control for lane keeping and obstacle avoidance on low curvature roads

This paper presents a control architecture based on a linear MPC formulation that addresses the lane keeping and obstacle avoidance problems for a passenger car driving on low curvature roads. The proposed control design decouples the longitudinal and lateral dynamics in two successive stages. First, plausible braking or throttle profiles are defined over the prediction horizon. Then, based on these profiles, linear time-varying models of the vehicle lateral dynamics are derived and used to formulate the associated linear MPC problems. The solutions of the optimization problems are used to determine for every time step, the optimal braking or throttle command and the corresponding steering angle command. Simulations show the ability of the controller to overcome multiple obstacles and keep the lane. Experimental results on an autonomous passenger vehicle driving on slippery roads show the effectiveness of the approach.

Predictive control of an autonomous ground vehicle using an iterative linearization approach

This paper presents the design of a controller for an autonomous ground vehicle. The goal is to track the lane centerline while avoiding collisions with obstacles. A nonlinear model predictive control (MPC) framework is used where the control inputs are the front steering angle and the braking torques at the four wheels. The focus of this work is on the development of a tailored algorithm for solving the nonlinear MPC problem. Hardware-in-the-loop simulations with the proposed algorithm show a reduction in the computational time as compared to general purpose nonlinear solvers. Experimental tests on a passenger vehicle at high speeds on low friction road surfaces show the effectiveness of the proposed algorithm.

A Learning-Based Framework for Velocity Control in Autonomous Driving

We present a framework for autonomous driving which can learn from human demonstrations, and we apply it to the longitudinal control of an autonomous car. Offline, we model car-following strategies from a set of example driving sequences. Online, the model is used to compute accelerations which replicate what a human driver would do in the same situation. This reference acceleration is tracked by a predictive controller which enforces a set of comfort and safety constraints before applying the final acceleration. The controller is designed to be robust to the uncertainty in the predicted motion of the preceding vehicle. In addition, we estimate the confidence of the driver model predictions and use it in the cost function of the predictive controller. As a result, we can handle cases where the training data used to learn the driver model does not provide sufficient information about how a human driver would handle the current driving situation. The approach is validated using a combination of simulations and experiments on our autonomous vehicle.

Driver Models for Personalised Driving Assistance

We propose a learning-based driver modelling approach which can identify manoeuvres performed by drivers on the highway and predict the future driver inputs. We show how this approach can be applied to provide personalised driving assistance. In a first example, the driver model is used to predict unintentional lane departures and a model predictive controller is used to keep the car in the lane. In a second example, the driver model estimates the preferred acceleration of the driver during lane keeping, and a model predictive controller is implemented to provide a personalised adaptive cruise control. For both applications, we use a combination of real data and simulation to evaluate the proposed approaches.

Autonomous car following: A learning-based approach

We propose a learning-based method for the longitudinal control of an autonomous vehicle on the highway. We use a driver model to generate acceleration inputs which are used as a reference by a model predictive controller. The driver model is trained using real driving data, so that it can reproduce the drivers behavior. We show the systems ability to reproduce different driving styles from different drivers. By solving a constrained optimization problem, the model predictive controller ensures that the control inputs applied to the vehicle satisfy some safety criteria. This is demonstrated on a vehicle by artificially creating potentially dangerous situations with virtual obstacles.

Manoeuvre generation and control for automated highway driving

A hierarchical, two-level architecture for manoeuvre generation and vehicle control for automated highway driving is presented. The high-level planner computes a manoeuvre in terms of a (X, Y)-trajectory as well as a longitudinal velocity profile, utilizing a simplified point-mass model and linear collision avoidance constraints. The low-level controller utilizes a non-linear vehicle model in order to compute the vehicle control inputs required to execute the planned manoeuvre. Both the high-level planner and low-level controller are formulated based on the model predictive control methodology. Simulation results demonstrates the ability of the high-level planner to compute appropriate, traffic-dependent manoeuvres, that can be tracked by the low-level controller in real-time.

Autonomous racing using learning Model Predictive Control

A novel learning Model Predictive Control technique is applied to the autonomous racing problem. The goal of the controller is to minimize the time to complete a lap. The proposed control strategy uses the data from previous laps to improve its performance while satisfying safety requirements. A system identification technique is proposed to estimate the vehicle dynamics. Simulation results with the high fidelity simulator software CarSim show the effectiveness of the proposed control scheme.

Stochastic predictive control for lane keeping assistance systems using a linear time-varying model

This paper presents a new controller for prevention of unintended roadway departures using model predictive control (MPC). The uncertainty with the drivers behavior is taken into account as the Gaussian disturbance. Correspondingly, we impose a lower bound on the probability of the vehicle remaining within the lane. Using current information of the vehicle and predicted steering of the driver, a linear time-varying (LTV) model of the human-vehicle system is obtained on-line through a successive linearization approach. The probabilistic safety constraints are converted into tightened constraints on the states of the LTV model by computing the evolution of the disturbance. Consequently, the predictive control problem is formulated as a quadratic program. The controller corrects the drivers steering, wherever there is a risk of unintended roadway departure, to keep the vehicle within the lane. Simulations and experiments implemented on a passenger vehicle were performed. The results indicate that the proposed controller improves safety performance compared to previous works.

Robust nonlinear predictive control for semiautonomous ground vehicles

This paper presents a robust control framework for lane-keeping and obstacle avoidance of semiautonomous ground vehicles. It presents a systematic way of enforcing robustness during the MPC design stage. A robust nonlinear Model Predictive Controller (RNMPC) is used to help the driver avoid obstacles and track the road center line. A force-input nonlinear bicycle vehicle model is developed for the RNMPC control design. A robust invariant set is used in the RNMPC design to ensure robust satisfaction of state and input constraints in the presence of disturbances and model errors. Simulations and experiments on testing vehicles show the effectiveness of the proposed framework.