Simulation research on emergency path planning of an active collision avoidance system combined with longitudinal control for an autonomous vehicle

Abstract

Autonomous vehicle technology is greatly valued nowadays, and an active collision avoidance system is one of the key parts for autonomous driving. This study presents a comprehensive architecture of an active collision avoidance system for an autonomous vehicle, which is integrated with a decision-making module, a path-planning module, a lateral-path-following module and a fuzzy adaptive following module (longitudinal motion) to deal with potential hazards on a straight road or a curved road. In order to make the planned path for overtaking manoeuvres safer, an improved harmonic velocity potential approach for path planning is presented, which innovatively enhances the effect of an obstacle potential on a road by adding a scale term, so that it can generate a smooth path for a vehicle-overtaking manoeuvre. All the potentials which are used for vehicle lane keeping or lane changing are well designed. The lateral-path-following module is based on the constrained linear model predictive control approach, which ensures that the host vehicle can follow the planned path precisely. Furthermore, when the overtaking manoeuvre is not suitable, the fuzzy adaptive following module is utilized to ensure that the host vehicle can adaptively keep a safe distance from the preceding vehicle. Tactical decisions, such as overtaking, accelerating or decelerating, are determined by the decision-making module. Finally, several typical scenarios with low traffic on a straight road or a curved road are simulated to verify the effectiveness and the feasibility of the active collision avoidance system. The simulation results show that the host vehicle can make a successful collision avoidance manoeuvre without the intervention of a human driver in different situations.

The Impact of Social Influence, Technophobia, and Perceived Safety on Autonomous Vehicle Technology Adoption

“…Due to the strongest correlation within this study, the perceived safety construct may serve as a significant focus area for the marketers of AVT. The consumer belief that AVT will enhance the safety of its passengers and the surroundings may lead to developing the trust and confidence required for the adoption of AVT (Bansal et al, 2016;Cao et al, 2016).…”

Longitudinal and lateral control of autonomous vehicles in multi‐vehicle driving environments

“…In response to static and dynamic obstacles around the vehicle, the trajectory planning generates lateral and longitudinal profiles for the vehicle to drive along the given path. For emergency path planning in active collision avoidance, Cao et al [16] presented an improved harmonic velocity potential approach for path planning, which strongly enhanced the effect of obstacle potentials on a road. Nilsson et al [17] formulated trajectory planning as the solution of a constrained optimal control problem over a finite time horizon, where additional constraints were introduced to avoid collisions with surrounding vehicles.…”

Shared steering control combined with driving intention for vehicle obstacle avoidance

“…Elastic band theory based on the artificial potential field method has been studied deeply for autonomous vehicle path planning. 24,25,37 In this section, the elastic band method considering driving intention for path planning is introduced for shared steering control.…”

“…The driver’s driving intention is recognized on-line by a support vector machine (SVM) model, and the recognition results are employed for path re-planning to match the driver’s driving intention. As shown in Figure 1, given the environment and obstacle information, the path to avoid obstacle is pre-planned by the elastic band algorithm off-line, which are studied well in our previous work by Song et al 24 and Cao et al 25 If a driver changes the current lane to avoid the obstacle, the path would be updated before the signal is fed to the path-following controller. The controller’s steering input is calculated by the well-studied MPC path-following algorithm.…”