Motion Planning Scott Kiesel Department of Computer Science skiesel at cs.unh.edu Scott Kiesel (UNH) Motion Planning – 1 / 14
Overview Motion Planning in Introduction ■ Overview 1. Discrete Space Discrete Space Continuous Space Grid with Smoothing Biasing Lattices 2. Continuous Space RRTs RRT* F-Biasing F-Biased Shell Scott Kiesel (UNH) Motion Planning – 2 / 14
Introduction Discrete Space ■ Discretization ■ Smoothing ■ Lattices Continuous Space Biasing Discrete Space Scott Kiesel (UNH) Motion Planning – 3 / 14
Discretization Introduction Discrete Space ■ Discretization ■ Smoothing ■ Lattices Continuous Space Biasing Optimal Complete Discrete Yes Yes Continuous No No Scott Kiesel (UNH) Motion Planning – 4 / 14
Smoothing Introduction 1. Discretize and apply A* Discrete Space ■ Discretization ■ Smoothing ■ Lattices Continuous Space 2. Apply post processing Biasing smoothing 3. Dynamically feasible path Scott Kiesel (UNH) Motion Planning – 5 / 14
Lattices Introduction Discrete Space ■ Discretization ■ Smoothing ■ Lattices Continuous Space Biasing Scott Kiesel (UNH) Motion Planning – 6 / 14
Introduction Discrete Space Continuous Space ■ RRTs ■ RRT* Biasing Continuous Space Scott Kiesel (UNH) Motion Planning – 7 / 14
RRTs Rapidly Exploring Random Trees (RRTs) Introduction Discrete Space Continuous Space ■ RRTs ■ RRT* Biasing Optimal Complete Discrete – – Continuous No Yes* Scott Kiesel (UNH) Motion Planning – 8 / 14
RRT* Introduction Discrete Space Continuous Space ■ RRTs (1) (2) (3) ■ RRT* Biasing (4) (5) (6) (7) Optimal Complete Discrete – – Continuous Yes* Yes* Scott Kiesel (UNH) Motion Planning – 9 / 14
Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell ■ Q&A Biasing Scott Kiesel (UNH) Motion Planning – 10 / 14
Simple Biasing Unbiased Random Sampling Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell ■ Q&A Scott Kiesel (UNH) Motion Planning – 11 / 14
Simple Biasing Goal Biased Sampling Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell ■ Q&A Scott Kiesel (UNH) Motion Planning – 11 / 14
F-Biasing Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell ■ Q&A First Solution RRT Anytime Profiles RRT goal 1% 0.8 goal 10% goal 25% 1600 f-biased best cost / cost unbiased cost 0.4 f-biased 1400 unbiased goal 1% goal 25% goal 10% 0 30 60 90 0 80 160 time (seconds) time (seconds) Scott Kiesel (UNH) Motion Planning – 12 / 14
Shell Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell Unbiased Goal-Biased F-Biased F-Biased Shell ■ Q&A Unbiased 20,000 Shell 2,500 Scott Kiesel (UNH) Motion Planning – 13 / 14
Q&A Introduction Discrete Space Continuous Space Biasing ■ Simple Biasing ■ F-Biasing ■ Shell ■ Q&A Questions? skiesel at cs.unh.edu Sound interesting? Come do research with us! Scott Kiesel (UNH) Motion Planning – 14 / 14
Recommend
More recommend