High-speed Autonomous Vehicle Research Articles

Nonlinear model predictive control for trajectory-planning and tracking based on tilting technology to achieve vehicle obstacle avoidance

ABSTRACT High-speed autonomous vehicles can effectively improve the performance of obstacle-avoidance trajectory-planning and tracking control through integration with existing electronic control systems. However, relatively little research has been conducted in this field. This paper proposes a novel methodology of obstacle-avoidance trajectory-planning and tracking control based on active front-wheel steering integrated with tilting technology that can tilt the vehicle body toward the inside of the curve via active suspension when turning. The controller is designed using hierarchical control, in which the upper layer uses the point-mass vehicle model to design the trajectory-planning algorithm based on model predictive control. The lower layer uses the nonlinear vehicle model to design obstacle-avoidance tracking nonlinear model predictive control based on active steering integrated with tilt control. Then, the constrained nonlinear model predictive control problem is transformed into a constrained nonlinear programming problem, which is solved by sequential quadratic programming. Finally, simulations were performed using the CarSim/Simulink co-simulation platform. Two other hierarchical obstacle-avoidance tracking nonlinear model predictive controllers were designed as comparison objects. The simulation results show that the planning trajectory of the proposed integrated controller is closest to the obstacle. This controller effectively improves the vehicle obstacle-avoidance trajectory-tracking performance, handling stability, and maneuverability.

Regional feature fusion for on-road detection of objects using camera and 3D-LiDAR in high-speed autonomous vehicles

Regional feature fusion for on-road detection of objects using camera and 3D-LiDAR in high-speed autonomous vehicles

Nonlinear Model Predictive Planning and Control for High-Speed Autonomous Vehicles on 3D Terrains

Nonlinear Model Predictive Planning and Control for High-Speed Autonomous Vehicles on 3D Terrains

Scenario Analysis and Control Comparison for a High Speed Autonomous Vehicle

<div class="section abstract"><div class="htmlview paragraph">This paper studies the simulation and control of an autonomous dragster. Four scenarios are provided that are critical to vehicle and driver safety in drag racing. Equations are then created to model the behavior during these safety scenarios. The use of a kinematic bicycle model and a Newtonian wheel stand model are discussed for plane-of-motion and out-of-plane vehicle movement, respectively. A separate controller is designed for each model by comparing different control methods. Proportional-Integral-Derivative (PID) control, optimal control, and model predictive control (MPC) are presented and applied to the models. The models are simulated from a speed of 75 m/s, being the estimated top speed of the research vehicle, up to a top speed of 150.5 m/s which is in alignment with the highest recorded speed of a dragster. The comparison of the control techniques yields MPC as superior for the bicycle model and PID as sufficient for the wheel stand model. Latency of the system is also discussed and accounted for in the system.</div></div>

Lane-following method for high speed autonomous vehicles

This paper presents a steering control method for lane-following in a vehicle using an image sensor. With each image frame acquired from the sensor, the steering control method determines target position and direction, and constructs a travel path from the current position to the target position either as an Arc-path or S-path. The steering angle is calculated from the travel path thus generated, and the vehicle follows the travel path via motor-control. The method was tested using a vehicle dubbed as KAV (Korea Autonomous Vehicle) along an expressway (Seoul Inner Beltway) trajectory with a variety of radii (50 m ∼ 300 m) while traveling at a speed of 60 km/h to 80 km/h. Compared with an experienced human driver, the method showed little much difference in performance in terms of lane-center deviation. The proposed method is currently employed for high speed autonomous driving as well as for stop and go traffic.