TY - GEN
T1 - Ground Plane Estimation for Obstacle Avoidance During Fixed-Wing UAV Landing
AU - Pęszor, Damian
AU - Wojciechowski, Konrad
AU - Szender, Marcin
AU - Wojciechowska, Marzena
AU - Paszkuta, Marcin
AU - Nowacki, Jerzy Paweł
N1 - Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021
Y1 - 2021
N2 - The automatic crash-landing of a fixed-wing UAV is challenging due to high velocity of approaching aircraft, limited manoeuvrability, lack of ability to hover and low quality of textural features in common landing sites. Available algorithms for ground plane estimation are designed for automatic cars, robots, and other land-bound platforms, where textural features can be easily compared due to low altitude, or for rotor-wing UAV that can hover. Their usefulness is limited when quick manoeuvre is needed to avoid collision with obstacles. Due to developments in parallelised, mobile computational architectures, an approach based on dense disparity estimation becomes available assuming proper constraints on ground plane transformation phase space. We propose an algorithm utilising such constraints for ground plane estimation on the often prohibitively time-consuming task of disparity calculation as well as plane estimation itself using a pyramid-based approach and random sample consensus in order to discard pixels belonging to obstacles. We use Inertial Navigation System, commonly available in fixed-wing UAVs, as a source of such constraints and improve our estimation in subsequent frames allowing for stable flight trajectory as well as detection of obstacles protruding from the ground plane.
AB - The automatic crash-landing of a fixed-wing UAV is challenging due to high velocity of approaching aircraft, limited manoeuvrability, lack of ability to hover and low quality of textural features in common landing sites. Available algorithms for ground plane estimation are designed for automatic cars, robots, and other land-bound platforms, where textural features can be easily compared due to low altitude, or for rotor-wing UAV that can hover. Their usefulness is limited when quick manoeuvre is needed to avoid collision with obstacles. Due to developments in parallelised, mobile computational architectures, an approach based on dense disparity estimation becomes available assuming proper constraints on ground plane transformation phase space. We propose an algorithm utilising such constraints for ground plane estimation on the often prohibitively time-consuming task of disparity calculation as well as plane estimation itself using a pyramid-based approach and random sample consensus in order to discard pixels belonging to obstacles. We use Inertial Navigation System, commonly available in fixed-wing UAVs, as a source of such constraints and improve our estimation in subsequent frames allowing for stable flight trajectory as well as detection of obstacles protruding from the ground plane.
KW - Disparity
KW - Fixed wing
KW - Ground plane estimation
KW - Obstacle avoidance
KW - UAV
KW - Unmanned aerial vehicle
UR - https://www.scopus.com/pages/publications/85104749003
U2 - 10.1007/978-3-030-73280-6_36
DO - 10.1007/978-3-030-73280-6_36
M3 - Conference contribution
AN - SCOPUS:85104749003
SN - 9783030732790
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 454
EP - 466
BT - Intelligent Information and Database Systems - 13th Asian Conference, ACIIDS 2021, Proceedings
A2 - Nguyen, Ngoc Thanh
A2 - Chittayasothorn, Suphamit
A2 - Niyato, Dusit
A2 - Trawiński, Bogdan
PB - Springer Science and Business Media Deutschland GmbH
T2 - 13th Asian Conference on Intelligent Information and Database Systems, ACIIDS 2021
Y2 - 7 April 2021 through 10 April 2021
ER -