Longitudinal Control Design for Autonomous HDVs

Abstract: Autonomous vehicles are evolving from a futuristic dream to current realityas technology becomes more and more mature. However, their presence on thehighways will demands precise control over lateral and longitudinal positioning.This thesis, completed at Scania Group AB in Södertälje, mainly focuses on thedesign of longitudinal controllers for trajectory tracking without compromisingalready mature control solutions. To this end, a virtual vehicle is created toextrapolate where the vehicle is supposed to be in time, and its position andvelocity are used as reference inputs. Then, two different control strategies (aPD controller and a lead compensator) based on position error and velocityerror are developed. A breaking algorithm is also developed to use servicebrakes on an Heavy-duty vehicle effectively. Numerical studies highlight thatthe proposed controllers exhibit excellent performance while tracking both positionand velocity trajectories. The proposed controllers are examined in threedifferent test cases: varying velocity, executing lane changes, and driving in anenvironment with significant changes in topology. In all test cases, the positionerror is less than 1m, while the velocity error is less than 0.5m/s. Comparedto the existing controller, developed in Scania Group AB, these controllers improvethe tracking performance significantly. Also, both controllers improverobustness against lateral changes and exogenous forces.