Autonomous Vehicle Maneuvering at the Limit of Friction

Abstract

Cover picture: The cover page picture is an illustrated version of a force-slip diagram showing the tire slips and forces visited during a maneuver. This is sometimes called a Nielsen diagram.

Lane-deviation penalty formulation and analysis for autonomous vehicle avoidance maneuvers

“…Another appealing approach is to apply receding-horizon control with LDP as objective in the optimization problem solved over the finite prediction horizon. Adopting a strategy such as in Fors 37 with a two-level architecture can then enable real-time control. There, an optimization problem with a low-complexity model is solved at the higher level, resulting in an acceleration profile that is subsequently refined in several steps considering more detailed vehicle dynamics to result in the actual actuator inputs in the lower level.…”

“…The desired optimal trajectories are computed as the solution to an optimization problem. In this optimization problem, the objective to be minimized is an integral of the sum of ( 34) and (36) as in (37), and to complete the problem formulation, model constraints on vehicle, obstacle, and road from Section ''Modeling'' are recollected together with initial and terminal conditions in (38)-( 45). The motion-planning problem for the studied double lane-change maneuver is consequently formulated as a continuous-time optimization problem in the following way: minimize…”

“…The obtained paths and selected trajectories for the model variables are shown in Figure 8(a). The same LDP objective (37) is used for all optimizations, with parameters Y o and Y 1 that are motivated by the road and the vehicle width, but to account for the obstacle geometry, the bottom drivable area is modified, namely Y bbp in (33). The characteristics of the XY-paths are very similar when comparing different widths of the obstacle.…”

Data-Driven Interaction-Aware Behavior Prediction for Autonomous Vehicles

“…In recent research, the use of predictive controllers is becoming increasingly prevalent, i.e., methods that encapsulate both the task of the motion planner and the feedback controller. Predictive controllers are commonly based on receding-horizon optimization, or more commonly, Model Predictive Control (MPC) (Batkovic, 2022, Fors, 2020, Ljungqvist, 2020, Zhou et al, 2022b. The MPC framework is an undeniably interesting application for the behavior prediction problem, in particular, because it conveniently allows for incorporating long-term predictions, offering the potential of developing proactive methods based on inter-agent interactions (Batkovic, 2022).…”

Fusion of Heterogeneous Friction Estimates for Traction Adaptive Motion Planning and Control

“…In critical situations, full utilization of the physical capacity of the vehicle is paramount to minimizing the risk of an accident, since overly cautious behavior will impair accident mitigation performance. This motivates research in motion planning and control at the limits of handling -a field that has seen a lot of progress in the last decade [1][2][3][4][5]. At the limit of tire adhesion, longitudinal, lateral and yaw dynamics are tightly coupled.…”