Simple Clothoid Lane Change Trajectories for Automated Vehicles Incorporating Friction Constraints

Abstract

This paper demonstrates that an autonomous vehicle can perform emergency lane changes up to the friction limits through real-time generation and evaluation of bi-elementary paths. Path curvature and friction determine the maximum possible speed along the path and, consequently, the feasibility of the path. This approach incorporates both steering inputs and changes in speed during the maneuver. As a result, varying path parameters and observing the maximum possible entry speed of resulting paths give insight about when and to what extent a vehicle should brake and turn during emergency lane change maneuvers. Tests on an autonomous vehicle validate this approach for lane changes near the limits of friction.

Importance of Vertical Dynamics for Accurate Modelling, Friction Estimation and Vehicle Motion Control

“…If the difference between the two signals starts to grow, this would mean that friction is no longer sufficient to generate the desired acceleration, and we can therefore update the control logic to avoid the loss of control of the vehicle. In fact, as it was reported in [1] and [2], the speed profile of a path should be generated while respecting the friction circle 2 , which can be represented as:…”

“…The vehicle should be able for example to detect an obstacle, update its planned trajectory, and control its dynamics to follow the desired trajectory. In [1], trajectories have been calculated using simple clothoids for lane change. What is more interesting about this research, is the fact that both the trajectory planning and the velocity profile generation are closely related to friction constraints.…”

Spatial Model Predictive Control for Smooth and Accurate Steering of an Autonomous Truck

“…Nevertheless, according to [28], in the past two decades, autonomous driving has triggered research that targets those issues. The problem of path planning has been addressed using mainly interpolation-based methods, such as clothoids [29], [30] and polynomial curves [31]. Among other approaches, we can refer to sampling-based methods, such as rapidly-exploring random trees (RRT) [32], lattice-based motion planners [33], [34], optimal control theory [35], and hybrid approaches [27].…”

Vehicle state estimation based on PSO-RBF neural network

“…The performance of driving safety systems may be improved if the unknown parameters of the underlying vehicle model can be measured and updated. Weight of the vehicle, road adhesion, drag coefficient and tyre cornering stiffness are examples of unknown parameters (Bektache et al, 2014;Funke and Gerdes, 2016;Rajaram and Subramanian, 2016;Chen et al, 2013a;Chen and Lee, 2011).…”