Reinforcement Learning in Mobile Robot Navigation: Literature Review

PDF

Authors
  1. Vincent, I.
Corporate Authors
Defence R&D Canada - Suffield, Ralston ALTA (CAN)
Abstract
Robotics is gaining popularity in military operations to facilitate soldier tasks and decrease their exposure to dangerous situations. Researchers are currently working on autonomous and semi-autonomous robots to provide soldiers with more intelligent robotic vehicles. At DRDC Suffield, the Autonomous Intelligent Systems Section is building an expertise in intelligent mobility for unmanned ground vehicles by developing robotic platforms that autonomously generate useful locomotive behaviours. One objective of the group is to solve the problem of navigability of shape-shifting mobile robot platforms. Learning to choose the appropriate geometric configuration with respect to the environment is a promising solution. This document presents a literature review emphasizing reinforcement learning in mobile robot navigation. A variety of algorithms are examined in detail such as Q-learning, HEDGER, SRV, RL-LSTM, CQ-L, HQ-L and W-learning. Finally, the applicability of those algorithms to the problem of shape-shifting mobile robot navigation is discussed.

Il y a un résumé en français ici.

Report Number
DRDC-SUFFIELD-TM-2006-117 — Technical Memorandum
Date of publication
01 Dec 2006
Number of Pages
38
DSTKIM No
CA029146
CANDIS No
527368
Format(s):
CD ROM

Permanent link

Document 1 of 1

Date modified: