Efficient Motion Planning for Autonomous Vehicle Maneuvers Using Duality-Based Decomposition

Forskningsoutput: TidskriftsbidragArtikel i vetenskaplig tidskrift

Abstract

A method to decompose a motion-planning problem into several segments is presented. It is based on a modification of the original problem, such that certain variables at the splitting points are considered to be precomputed and thus fixed and the remaining variables are obtained by performing Lagrange relaxation. The resulting dual problem is split into several subproblems, allowing parallel computation. The method is formalized as a computational algorithm and evaluated in a safety critical double lane-change situation. The resulting maneuver has close-to-optimal behavior and, for certain initialization strategies, it is obtained in shorter computational time compared to computing the full maneuver in one step.

Detaljer

Författare
Enheter & grupper
Externa organisationer
  • Linköping University
Forskningsområden

Ämnesklassifikation (UKÄ) – OBLIGATORISK

  • Beräkningsmatematik
  • Datavetenskap (datalogi)

Nyckelord

Originalspråkengelska
Sidor (från-till)78-84
Antal sidor7
TidskriftIFAC-PapersOnLine
Volym52
Utgåva nummer5
StatusPublished - 2019 sep 19
PublikationskategoriForskning
Peer review utfördJa
Evenemang9th IFAC Symposium on Advances in Automotive Control, AAC 2019 - Orléans, Frankrike
Varaktighet: 2019 jun 232019 jun 27