Abstract:

Aiming to tackle the computation challenges arising from the stringent positioning accuracy requirement as well as the uncertainties in system dynamics, this paper presents a fast adaptive planning method for tractor-trailer systems.
Building on the delayed-expansion A-search guided tree (DEAGT) that grows a tree from pre-computed motion primitives
(MPs) via A*-like search, Adaptive-iAGT contributes: (i) an iterative Linear-Quadratic Regulator (iLQR) steering module that reaches the goal without a reference trajectory, reducing the planning time; (ii) a two-stage framework that resolves parameter uncertainties by adapting a nominal trajectory to true parameters via iLQR, eliminating the need to store MPs for various configurations; and (iii) analytical control scaling that lets iLQR ignore velocity/control bounds and reinstates them afterwards, increasing the success rate. Simulations and field tests validate the effectiveness of the proposed algorithm.