Convex relaxation techniques and model predictive control for autonomous driving of heavy trucks

Many problems of interest in systems and control can be formulated in the framework of convex optimization by using convex relaxation techniques. A well-known example is the use of the l_1-norm as a substitute for the so-called l_0-norm in order to enhance sparseness in a parameter vector. We will discuss how to  use the l_1- norm, with extensions, to handle piecewise constant or linear constraints. This technique is used in mean and trend filtering for change point detection.  We will apply this approach to model predictive control of self-driving heavy trucks.  The control objective is based on the minimization of the second and first spatial derivatives of the curvature.  We ensure good path tracking accuracy by adding constraints that avoid deviations from the reference path. This leads to smooth and accurate driving. We will also apply this concept to optimal speed profiles for given vehicle limitations and road constraints. The algorithms are implemented and evaluated on a Scania construction truck.
 
Bio: Bo Wahlberg received the M.Sc. degree in Electrical Engineering in 1983 and the Ph.D. degree in 1987 from Linköping University, Sweden. He was a post-doc at University of Newcastle, Australia, in 1988. In December 1991, he became Professor of the Chair of Automatic Control at KTH Royal Institute of Technology, Stockholm, Sweden. He was a visiting professor at the Department of Electrical Engineering, Stanford University, USA, August 1997 - July 1998 and August 2009 -June 2010, and vice president of KTH 1999 - 2001. He is a Fellow of the IEEE for his contributions to system identification using orthonormal basis functions.
Bo Wahlberg is a co-founder of Centre of Autonomous Systems and the Linnaeus Center ACCESS on networked systems at KTH. He is the KTH director for the Wallenberg Autonomous Systems Program (WASP) , that recently was granted over 200 million USD over ten years for research into autonomous systems and software development. His research interests include system identification, modeling and control of industrial processes, and statistical  signal processing with applications in smart transportation systems.