Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning

Brian Gaudet, Roberto Furfaro

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

Future unconstrained and science-driven missions to Mars will require advanced guidance algorithms that are able to adapt to more demanding mission requirements, e.g. landing on selected locales with pinpoint accuracy while autonomously flying fuel-efficient trajectories. In this paper, we will present a novel guidance algorithm designed by applying the principles of Reinforcement Learning (RL) theory. The goal is to devise an adaptive guidance algorithm that enables robust, fuel efficient, and accurate landing without the need for off-line trajectory generation. Results from a Monte Carlo simulation campaign show that the algorithm is capable of autonomously flying trajectories that are close to the optimal minimum-fuel solutions with an accuracy that surpasses conventional Apollo-like guidance algorithms. The proposed RL-based guidance algorithm exhibits a high degree of flexibility and can easily accommodate autonomous retargeting while maintaining accuracy and fuel efficiency. Although reinforcement learning and other similar machine learning techniques have been previously applied to aerospace guidance and control problems (e.g., autonomous helicopter control), this appears, to the best of our knowledge, to be the first application of reinforcement learning to the problem of autonomous planetary landing.

Original languageEnglish (US)
Title of host publicationAdvances in the Astronautical Sciences
Pages1309-1328
Number of pages20
Volume143
StatePublished - 2012
Event22nd AAS/AIAA Space Flight Mechanics Meeting - Charleston, SC, United States
Duration: Feb 2 2012Feb 2 2012

Other

Other22nd AAS/AIAA Space Flight Mechanics Meeting
CountryUnited States
CityCharleston, SC
Period2/2/122/2/12

Fingerprint

Mars landing
Reinforcement learning
reinforcement
Landing
learning
Mars
landing
trajectory
Trajectories
trajectories
helicopter control
planetary landing
learning theory
flight
machine learning
Helicopters
mars
Learning systems
flexibility
requirements

ASJC Scopus subject areas

  • Aerospace Engineering
  • Space and Planetary Science

Cite this

Gaudet, B., & Furfaro, R. (2012). Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning. In Advances in the Astronautical Sciences (Vol. 143, pp. 1309-1328)

Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning. / Gaudet, Brian; Furfaro, Roberto.

Advances in the Astronautical Sciences. Vol. 143 2012. p. 1309-1328.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Gaudet, B & Furfaro, R 2012, Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning. in Advances in the Astronautical Sciences. vol. 143, pp. 1309-1328, 22nd AAS/AIAA Space Flight Mechanics Meeting, Charleston, SC, United States, 2/2/12.
Gaudet B, Furfaro R. Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning. In Advances in the Astronautical Sciences. Vol. 143. 2012. p. 1309-1328
Gaudet, Brian ; Furfaro, Roberto. / Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning. Advances in the Astronautical Sciences. Vol. 143 2012. pp. 1309-1328
@inproceedings{29f128444b904bec905c13b8df09bc8d,
title = "Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning",
abstract = "Future unconstrained and science-driven missions to Mars will require advanced guidance algorithms that are able to adapt to more demanding mission requirements, e.g. landing on selected locales with pinpoint accuracy while autonomously flying fuel-efficient trajectories. In this paper, we will present a novel guidance algorithm designed by applying the principles of Reinforcement Learning (RL) theory. The goal is to devise an adaptive guidance algorithm that enables robust, fuel efficient, and accurate landing without the need for off-line trajectory generation. Results from a Monte Carlo simulation campaign show that the algorithm is capable of autonomously flying trajectories that are close to the optimal minimum-fuel solutions with an accuracy that surpasses conventional Apollo-like guidance algorithms. The proposed RL-based guidance algorithm exhibits a high degree of flexibility and can easily accommodate autonomous retargeting while maintaining accuracy and fuel efficiency. Although reinforcement learning and other similar machine learning techniques have been previously applied to aerospace guidance and control problems (e.g., autonomous helicopter control), this appears, to the best of our knowledge, to be the first application of reinforcement learning to the problem of autonomous planetary landing.",
author = "Brian Gaudet and Roberto Furfaro",
year = "2012",
language = "English (US)",
isbn = "9780877035817",
volume = "143",
pages = "1309--1328",
booktitle = "Advances in the Astronautical Sciences",

}

TY - GEN

T1 - Adaptive pinpoint and fuel efficient Mars landing using Reinforcement Learning

AU - Gaudet, Brian

AU - Furfaro, Roberto

PY - 2012

Y1 - 2012

N2 - Future unconstrained and science-driven missions to Mars will require advanced guidance algorithms that are able to adapt to more demanding mission requirements, e.g. landing on selected locales with pinpoint accuracy while autonomously flying fuel-efficient trajectories. In this paper, we will present a novel guidance algorithm designed by applying the principles of Reinforcement Learning (RL) theory. The goal is to devise an adaptive guidance algorithm that enables robust, fuel efficient, and accurate landing without the need for off-line trajectory generation. Results from a Monte Carlo simulation campaign show that the algorithm is capable of autonomously flying trajectories that are close to the optimal minimum-fuel solutions with an accuracy that surpasses conventional Apollo-like guidance algorithms. The proposed RL-based guidance algorithm exhibits a high degree of flexibility and can easily accommodate autonomous retargeting while maintaining accuracy and fuel efficiency. Although reinforcement learning and other similar machine learning techniques have been previously applied to aerospace guidance and control problems (e.g., autonomous helicopter control), this appears, to the best of our knowledge, to be the first application of reinforcement learning to the problem of autonomous planetary landing.

AB - Future unconstrained and science-driven missions to Mars will require advanced guidance algorithms that are able to adapt to more demanding mission requirements, e.g. landing on selected locales with pinpoint accuracy while autonomously flying fuel-efficient trajectories. In this paper, we will present a novel guidance algorithm designed by applying the principles of Reinforcement Learning (RL) theory. The goal is to devise an adaptive guidance algorithm that enables robust, fuel efficient, and accurate landing without the need for off-line trajectory generation. Results from a Monte Carlo simulation campaign show that the algorithm is capable of autonomously flying trajectories that are close to the optimal minimum-fuel solutions with an accuracy that surpasses conventional Apollo-like guidance algorithms. The proposed RL-based guidance algorithm exhibits a high degree of flexibility and can easily accommodate autonomous retargeting while maintaining accuracy and fuel efficiency. Although reinforcement learning and other similar machine learning techniques have been previously applied to aerospace guidance and control problems (e.g., autonomous helicopter control), this appears, to the best of our knowledge, to be the first application of reinforcement learning to the problem of autonomous planetary landing.

UR - http://www.scopus.com/inward/record.url?scp=84879398490&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84879398490&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9780877035817

VL - 143

SP - 1309

EP - 1328

BT - Advances in the Astronautical Sciences

ER -