Skip to content

Latest commit

 

History

History
21 lines (17 loc) · 849 Bytes

gameSLAM.md

File metadata and controls

21 lines (17 loc) · 849 Bytes

GTP-SLAM: Game-Theoretic Priors for Simultaneous Localization and Mapping in Multi-Agent Scenarios

Author:

Year: 2022

Notes:

  • SLAM as a game with an optimal solution
  • usefull in high dynamic scenarios (traffic)
  • this uses an iterative best response (IBR) scheme, in which each player takes a turn solving for their optimal strategy while assuming all other players’ strategies are fixed
  • IBR can be solved with a factor graph framework

Preliminaries:

  • there are $N$ players: one ego player and the others
  • jointly estimate trajectories off ego player and non-ego playerS
  • the ego player observes $N_l$ landmarks and players observe other players positions
  • Nash Equilibrium : a set of control input reaches a Nash Equilibrium if no player can lower their cost by deviating their input, the other players input being fixed