Randomized Bidirectional B-Spline Parameterization Motion Planning

Mohamed Elbanhawi

doi:10.1109/TITS.2015.2477355

Return

Randomized Bidirectional B-Spline Parameterization Motion Planning

2016-02-01

PRE

Mohamed Elbanhawi

Milan Simić

Reza N. Jazar *

DOI：10.1109/TITS.2015.2477355

Original

Original request for help

Save

Abstract

En 中文

This paper presents a motion planner tailored for particular requirements for robotic car navigation. We leverage B-spline curve properties to include vehicle's constraint requirements, thus lowering the search dimensionality. An algorithm, which combines competent exploratory nature of the randomized search methods with vector-valued parameterization steering, is developed here. Vehicle's limitations, along with obstacle's constraints, are satisfied without being hindered by numerical integration and control space discretization of traditional randomized kinodynamic planners. We rely on newly developed theoretical underpinnings to overcome performance issues in rapidly exploring random tree (RRT) solutions. Rigorous simulations and analysis demonstrate that this new approach outperforms recently proposed planners by using an efficient bidirectional RRT-based search, by maintaining continuous state and control spaces, and generating C-2 continuous paths, which are realistic inputs suited for mobile robotic applications and passenger vehicles.

Keywords:

RRT

nonholonomic planning

motion planning

continuous trajectory

B-spline path

autonomous vehicles

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.