Visual Odometry and SLAM using Line Segment Features Ruben - - PowerPoint PPT Presentation

▶
visual odometry and slam using line segment features
SMART_READER_LITE
LIVE PREVIEW

Visual Odometry and SLAM using Line Segment Features Ruben - - PowerPoint PPT Presentation

Apr 03, 2023 268 likes •590 views

Visual Odometry and SLAM using Line Segment Features Ruben Gomez-Ojeda Machine Perception and Intelligent Robotics (MAPIR), University of Malaga http://mapir.isa.uma.es Ou Outline I. Motivation II. Contributions to Visual Odometry i.

slide-1
SLIDE 1

Visual Odometry and SLAM using Line Segment Features

Ruben Gomez-Ojeda

Machine Perception and Intelligent Robotics (MAPIR), University of Malaga http://mapir.isa.uma.es

slide-2
SLIDE 2

I. Motivation II. Contributions to Visual Odometry i. Stereo VO ii. Monocular VO

  • III. Contributions to Visual SLAM
  • IV. Future Work

Ou Outline

slide-3
SLIDE 3

I.

  • I. Mo

Motivation

slide-4
SLIDE 4

Mo Motivation

  • R. Mur-Artal, J.M.M. Montiel, & J.D. Tardos (2015). OR

ORB-SL SLAM: a versatile and accurate monocular SL SLAM sy syst

  • stem. IEEE Transactions on Robotics.
  • R. Mur-Artal & J.D. Tardos (2017). OR

ORB-SL SLAM2: An An open-so source sl slam system for monocular, stereo, and rg rgb-d d ca camera

  • ras. IEEE Transactions on Robotics.

Despite its high accuracy, point-based VO approaches such as ORB-SLAM2 can lose the tracking in low-textured or bad illuminated scenarios, as the number of features decreases

slide-5
SLIDE 5

Mo Motivation

  • R. Mur-Artal, J.M.M. Montiel, & J.D. Tardos (2015). OR

ORB-SL SLAM: a versatile and accurate monocular SL SLAM sy syst

  • stem. IEEE Transactions on Robotics.
  • R. Mur-Artal & J.D. Tardos (2017). OR

ORB-SL SLAM2: An An open-so source sl slam system for monocular, stereo, and rg rgb-d d ca camera

  • ras. IEEE Transactions on Robotics.
slide-6
SLIDE 6

II.

  • II. Con

Contribution

  • ns to

to Visual Odo Odometry i. . St Stereo VO VO

  • R. Gomez-Ojeda & J. Gonzalez-Jimenez (2016). Ro

Robust Stereo Visual Od Odometry th through a Probabilisti tic Combinati tion of Po Points and Line Segments. IEEE International Conference on Robotics and Automation (ICRA), Stockholm, Sweden, 2016.

slide-7
SLIDE 7

St Stereo VO VO

  • R. G. Von Gioi, J. Jakubowicz, J.M. Morel & G. Randall (2010). LS

LSD: A fast st line se segment detector with a false se detection co

  • control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010.
  • L. Zhang & R. Koch (2013). An

An efficient and robust line segment matching approach based on LBD descriptor and pairwise ge geometric cons nsistenc

  • ncy. Journal of Visual Communication and Image Representation, 2013.
  • E. Rublee, V. Rabaud, K. Konolige & G. Bradski(2011). OR

ORB: An efficient alternative ve to SIFT or SUR

  • URF. IEEE Conference on

Computer Vision (ICCV), Barcelona, Spain, 2011.

OR ORB po poin ints:

  • Very efficient
  • Good performance
  • Only best mutual matches

LS LSD+ D+LB LBD D line se segme ments:

  • High precision and repeatability
  • Time consuming (modified version)
  • Only best mutual matches
  • Stereo projection of the end-points
  • Deal with partial occlusions

Ov Over erview:

  • Frame to frame approach
  • Robust performance in most scenarios
  • Probabilistic SE(3) minimization of

reprojection errors

slide-8
SLIDE 8

St Stereo VO VO

slide-9
SLIDE 9

ORB points LSD line segments

St Stereo VO VO

slide-10
SLIDE 10

St Stereo VO VO: : KI KITTI seq equen ences es

slide-11
SLIDE 11

KITTI-00 EuRoC/V1_01_easy 5x 2x

St Stereo VO VO

ht https://github.com/rubengooj/StVO-PL PL Source code available:

slide-12
SLIDE 12

II.

  • II. Con

Contribution

  • ns to

to Visual Odo Odometry ii.

  • ii. Monocula

lar VO

  • R. Gomez-Ojeda, J. Briales, & J. Gonzalez-Jimenez. PL

PL-SV SVO: : Semi-Di Direct Monocular Visual Od Odometry by Combining Points and Li Line ne Se

  • Segments. Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International Conference on, Daejeon, Korea, 2016
slide-13
SLIDE 13

Mo MonoVO

  • R. Gomez-Ojeda, J. Briales, & J. Gonzalez-Jimenez. PL

PL-SV SVO: : Semi-Di Direct t Monocular Vi Visual Odometr try by Co Combining P Points a and L Line Se

  • Segments. Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International

Conference on, Daejeon, Korea, 2016

Pros Cons Point-based Fast to detect and match Presence of outliers Usually abundant Reduction in structured scenarios Line-based Good behavior in most scenarios Slow detection and matching Less outliers (more distinctive feat.) Tracking of endpoints and occlusions Our approach SVO approach allow for the fast tracking of line segments Points along the line usually are not key- points, so the feature alignment step is less reliable Robust performance in both kind

  • f scenarios

Depth estimation of endpoints might be affected by occlusions

slide-14
SLIDE 14

SV SVO [1]: PL PL-SV SVO:

Figure adapted from [SVO]

Mo MonoVO: : Sp Sparse Model-ba based d Image Alignm nment

  • C. Forster, M. Pizzoli & D. Scaramuzza (2014). SV

SVO: Fast semi-di direct mono nocul ular ar od

  • dome
  • metry.

. IEEE International Conference on Robotics and Automation (ICRA), Hong Kong, China, 2014.

Seek: rigid body transformation between frames using current model

  • Direct approach, min photometric error
  • 4x4 patches around points are used
  • Coarse-to-fine scheme

Segment regions are not small:

  • Warping
  • Point sampling
slide-15
SLIDE 15

Seek: new feature positions

  • Min photometric error wrt corresponding path in

closest KF

  • 8x8 patch with affine warping

Fe Feature re refinement (S (SVO O [1], PL PL-SV SVO):

Figure adapted from [SVO]

Mo MonoVO: : In Individual Feature Alignment

  • C. Forster, M. Pizzoli & D. Scaramuzza (2014). SV

SVO: Fast semi-di direct mono nocul ular ar od

  • dome
  • metry.

. IEEE International Conference on Robotics and Automation (ICRA), Hong Kong, China, 2014.

Refine feature correspondences Segment features:

  • Use endpoints only (simplicity)
  • We deal with outliers in the final refinement step.
slide-16
SLIDE 16

Fast bundle adjustment Re Reprojection er errors: Mo Motion-on

  • nly BA

BA:

Figure adapted from [SVO]

Mo MonoVO: : Po Pose and Structure Refinement

  • C. Forster, M. Pizzoli & D. Scaramuzza (2014). SV

SVO: Fast semi-di direct mono nocul ular ar od

  • dome
  • metry.

. IEEE International Conference on Robotics and Automation (ICRA), Hong Kong, China, 2014.

Seek: poses and map

  • Min reprojection error
  • Reduces drift

Robustified framework:

  • Heuristics may introduce outliers
  • Cauchy loss function & outlier filter
slide-17
SLIDE 17

EuRoC/MH_05_difficult Low-tectured scene 1x 1x

Mo MonoVO

PL PL-SV SVO:

  • More robust performance
  • Still real-time: 60 Hz
  • Fast tracking and mapping of line segments

ht https://gi github. thub.com/rube /rubeng ngooj/pl j/pl-sv svo Source code available:

slide-18
SLIDE 18

III.

  • III. Con

Contribution

  • ns to

to Visual SLAM

  • R. Gomez-Ojeda, F.A. Moreno, D. Scaramuzza & J. Gonzalez-Jimenez (2017). PL

PL-SL SLAM: : a Stereo SLAM System through the Co Combination of Points and Line Se

  • Segments. ArXiv preprint arXiv:1705.09479, 2017.
slide-19
SLIDE 19

LB LBA:

St Stereo Vi Visua ual SLAM

slide-20
SLIDE 20

LB LBA:

St Stereo Vi Visua ual SLAM

slide-21
SLIDE 21

LC LC:

St Stereo Vi Visua ual SLAM

slide-22
SLIDE 22

Mapping example in EuRoC/V1_01_easy Low-textured scene 0.5x 5x

St Stereo SLA SLAM

ht https://gi github. thub.com/rube /rubeng ngooj/pl j/pl-sl slam am Source code available:

slide-23
SLIDE 23

IV

  • IV. Fut

Futur ure Wo Work

  • R. Gomez-Ojeda & J. Gonzalez-Jimenez (2017). Fa

Fast Li Line ne Segm gment nt Mat atchi hing ng for Stereo Visual sual Od Odometry. . Submitted to ICRA 2018.

slide-24
SLIDE 24

Pe Performance Co Compariso ison be between PL PLVO and and F-PL PLVO

Da Dataset: : EuRoC Se Sequence ce: V1_01_easy Mo Motion ty type: MAV Te Texture ty type: indoor, well-structured Il Illumination: : well-illuminated

slide-25
SLIDE 25

PLVO F-PLVO 5x 5x

slide-26
SLIDE 26

Pe Performance Co Compariso ison be between PL PLVO and and F-PL PLVO

Da Dataset: : EuRoC Se Sequence ce: V1_02_medium Mo Motion ty type: MAV (fast motion) Te Texture ty type: indoor, low-textured (partially) Il Illumination: : well-illuminated

slide-27
SLIDE 27

PLVO F-PLVO 5x 5x

slide-28
SLIDE 28

Pe Performance Co Compariso ison be between PL PLVO and and F-PL PLVO

Da Dataset: : Tsukuba Se Sequence ce: Lamps Mo Motion ty type: Synthetic Te Texture ty type: indoor, textured Il Illumination: : low-illuminated

slide-29
SLIDE 29

PLVO F-PLVO 3x 3x

slide-30
SLIDE 30

Pe Performance Co Compariso ison be between PL PLVO and and F-PL PLVO

Da Dataset: : EuRoC Se Sequence ce: MH_05_difficult Mo Motion ty type: MAV Te Texture ty type: indoor, well-structured Il Illumination: : changes of illumination

slide-31
SLIDE 31

PLVO F-PLVO 5x 5x

slide-32
SLIDE 32

Visual Odometry and SLAM using Line Segment Features

Ruben Gomez-Ojeda

Machine Perception and Intelligent Robotics (MAPIR), University of Malaga http://mapir.isa.uma.es