Image schema based landing and navigation for rotorcraft MAV-S

Eniko T Enikov, Juan Antonio Escareno, Micky Rakotondrabe

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

Abstract

To date, most autonomous micro air vehicles (MAV-s) operate in a controlled environment, where the location of and attitude of the aircraft are measured with an infrared (IR) tracking systems. If MAV-s are to ever exit the lab, their flight control needs to become autonomous and based on on-board image and attitude sensors. To address this need, several groups are developing monocular and binocular image based navigation systems. One of the challenges of these systems is the need for exact calibration in order to determine the vehicle's position and attitude through the solution of an inverse problem. Body schemas are a biologically-inspired approach, emulating the plasticity of the animal brain, which allows it to learn non-linear mappings between the body configurations, i.e. its generalized coordinates and the resulting sensory outputs. The advantages of body schemas has long been recognized in the cognitive robotic literature and resulting studies on human-robot interactions based on artificial neural networks, however little effort has been made so far to develop avian-inspired flight control strategies utilizing body and image schemas. This paper presents a numerical experiment of controlling the trajectory of a miniature rotorcraft during landing maneuvers suing the notion of body and image schemas. More specifically, we demonstrate how trajectory planning can be executed in the image space using gradient-based maximum seeking algorithm of a pseudo-potential. It is demonstrated that a neural-gas type artificial neural network (ANN), trained through Hebbian-type learning algorithm, can be effective in learning a mapping between the rotorcraft's position/attitude and the output of its vision sensors. Numerical simulation of the landing performance, including resulting landing errors are presented using an experimentally validated rotorcraft model.

Original languageEnglish (US)
Title of host publicationAdvances in Aerospace Technology
PublisherAmerican Society of Mechanical Engineers (ASME)
Volume1-2015
ISBN (Electronic)9780791857342
DOIs
StatePublished - 2015
EventASME 2015 International Mechanical Engineering Congress and Exposition, IMECE 2015 - Houston, United States
Duration: Nov 13 2015Nov 19 2015

Other

OtherASME 2015 International Mechanical Engineering Congress and Exposition, IMECE 2015
CountryUnited States
CityHouston
Period11/13/1511/19/15

Fingerprint

Micro air vehicle (MAV)
Landing
Navigation
Trajectories
Neural networks
Binoculars
Human robot interaction
Sensors
Navigation systems
Inverse problems
Learning algorithms
Plasticity
Brain
Animals
Robotics
Aircraft
Calibration
Infrared radiation
Planning
Computer simulation

ASJC Scopus subject areas

  • Mechanical Engineering

Cite this

Enikov, E. T., Escareno, J. A., & Rakotondrabe, M. (2015). Image schema based landing and navigation for rotorcraft MAV-S. In Advances in Aerospace Technology (Vol. 1-2015). American Society of Mechanical Engineers (ASME). https://doi.org/10.1115/IMECE2015-51450

Image schema based landing and navigation for rotorcraft MAV-S. / Enikov, Eniko T; Escareno, Juan Antonio; Rakotondrabe, Micky.

Advances in Aerospace Technology. Vol. 1-2015 American Society of Mechanical Engineers (ASME), 2015.

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

Enikov, ET, Escareno, JA & Rakotondrabe, M 2015, Image schema based landing and navigation for rotorcraft MAV-S. in Advances in Aerospace Technology. vol. 1-2015, American Society of Mechanical Engineers (ASME), ASME 2015 International Mechanical Engineering Congress and Exposition, IMECE 2015, Houston, United States, 11/13/15. https://doi.org/10.1115/IMECE2015-51450
Enikov ET, Escareno JA, Rakotondrabe M. Image schema based landing and navigation for rotorcraft MAV-S. In Advances in Aerospace Technology. Vol. 1-2015. American Society of Mechanical Engineers (ASME). 2015 https://doi.org/10.1115/IMECE2015-51450
Enikov, Eniko T ; Escareno, Juan Antonio ; Rakotondrabe, Micky. / Image schema based landing and navigation for rotorcraft MAV-S. Advances in Aerospace Technology. Vol. 1-2015 American Society of Mechanical Engineers (ASME), 2015.
@inproceedings{a6cb7cf327e2440084f40c4e9dd133f5,
title = "Image schema based landing and navigation for rotorcraft MAV-S",
abstract = "To date, most autonomous micro air vehicles (MAV-s) operate in a controlled environment, where the location of and attitude of the aircraft are measured with an infrared (IR) tracking systems. If MAV-s are to ever exit the lab, their flight control needs to become autonomous and based on on-board image and attitude sensors. To address this need, several groups are developing monocular and binocular image based navigation systems. One of the challenges of these systems is the need for exact calibration in order to determine the vehicle's position and attitude through the solution of an inverse problem. Body schemas are a biologically-inspired approach, emulating the plasticity of the animal brain, which allows it to learn non-linear mappings between the body configurations, i.e. its generalized coordinates and the resulting sensory outputs. The advantages of body schemas has long been recognized in the cognitive robotic literature and resulting studies on human-robot interactions based on artificial neural networks, however little effort has been made so far to develop avian-inspired flight control strategies utilizing body and image schemas. This paper presents a numerical experiment of controlling the trajectory of a miniature rotorcraft during landing maneuvers suing the notion of body and image schemas. More specifically, we demonstrate how trajectory planning can be executed in the image space using gradient-based maximum seeking algorithm of a pseudo-potential. It is demonstrated that a neural-gas type artificial neural network (ANN), trained through Hebbian-type learning algorithm, can be effective in learning a mapping between the rotorcraft's position/attitude and the output of its vision sensors. Numerical simulation of the landing performance, including resulting landing errors are presented using an experimentally validated rotorcraft model.",
author = "Enikov, {Eniko T} and Escareno, {Juan Antonio} and Micky Rakotondrabe",
year = "2015",
doi = "10.1115/IMECE2015-51450",
language = "English (US)",
volume = "1-2015",
booktitle = "Advances in Aerospace Technology",
publisher = "American Society of Mechanical Engineers (ASME)",

}

TY - GEN

T1 - Image schema based landing and navigation for rotorcraft MAV-S

AU - Enikov, Eniko T

AU - Escareno, Juan Antonio

AU - Rakotondrabe, Micky

PY - 2015

Y1 - 2015

N2 - To date, most autonomous micro air vehicles (MAV-s) operate in a controlled environment, where the location of and attitude of the aircraft are measured with an infrared (IR) tracking systems. If MAV-s are to ever exit the lab, their flight control needs to become autonomous and based on on-board image and attitude sensors. To address this need, several groups are developing monocular and binocular image based navigation systems. One of the challenges of these systems is the need for exact calibration in order to determine the vehicle's position and attitude through the solution of an inverse problem. Body schemas are a biologically-inspired approach, emulating the plasticity of the animal brain, which allows it to learn non-linear mappings between the body configurations, i.e. its generalized coordinates and the resulting sensory outputs. The advantages of body schemas has long been recognized in the cognitive robotic literature and resulting studies on human-robot interactions based on artificial neural networks, however little effort has been made so far to develop avian-inspired flight control strategies utilizing body and image schemas. This paper presents a numerical experiment of controlling the trajectory of a miniature rotorcraft during landing maneuvers suing the notion of body and image schemas. More specifically, we demonstrate how trajectory planning can be executed in the image space using gradient-based maximum seeking algorithm of a pseudo-potential. It is demonstrated that a neural-gas type artificial neural network (ANN), trained through Hebbian-type learning algorithm, can be effective in learning a mapping between the rotorcraft's position/attitude and the output of its vision sensors. Numerical simulation of the landing performance, including resulting landing errors are presented using an experimentally validated rotorcraft model.

AB - To date, most autonomous micro air vehicles (MAV-s) operate in a controlled environment, where the location of and attitude of the aircraft are measured with an infrared (IR) tracking systems. If MAV-s are to ever exit the lab, their flight control needs to become autonomous and based on on-board image and attitude sensors. To address this need, several groups are developing monocular and binocular image based navigation systems. One of the challenges of these systems is the need for exact calibration in order to determine the vehicle's position and attitude through the solution of an inverse problem. Body schemas are a biologically-inspired approach, emulating the plasticity of the animal brain, which allows it to learn non-linear mappings between the body configurations, i.e. its generalized coordinates and the resulting sensory outputs. The advantages of body schemas has long been recognized in the cognitive robotic literature and resulting studies on human-robot interactions based on artificial neural networks, however little effort has been made so far to develop avian-inspired flight control strategies utilizing body and image schemas. This paper presents a numerical experiment of controlling the trajectory of a miniature rotorcraft during landing maneuvers suing the notion of body and image schemas. More specifically, we demonstrate how trajectory planning can be executed in the image space using gradient-based maximum seeking algorithm of a pseudo-potential. It is demonstrated that a neural-gas type artificial neural network (ANN), trained through Hebbian-type learning algorithm, can be effective in learning a mapping between the rotorcraft's position/attitude and the output of its vision sensors. Numerical simulation of the landing performance, including resulting landing errors are presented using an experimentally validated rotorcraft model.

UR - http://www.scopus.com/inward/record.url?scp=84982987713&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84982987713&partnerID=8YFLogxK

U2 - 10.1115/IMECE2015-51450

DO - 10.1115/IMECE2015-51450

M3 - Conference contribution

VL - 1-2015

BT - Advances in Aerospace Technology

PB - American Society of Mechanical Engineers (ASME)

ER -