Vehicle Localization based on Lane Marking Detection Yuncong Chen - - PowerPoint PPT Presentation

▶
vehicle localization based on lane marking detection
SMART_READER_LITE
LIVE PREVIEW

Vehicle Localization based on Lane Marking Detection Yuncong Chen - - PowerPoint PPT Presentation

Sep 22, 2022 191 likes •564 views

Vehicle Localization based on Lane Marking Detection Yuncong Chen UCSD HRI intern 2014 Overview Input Goal Odometry lateral localization on highway (noisy GPS / IMU for now) give correct estimate on merge / split points

slide-1
SLIDE 1

Vehicle Localization based on Lane Marking Detection

Yuncong Chen UCSD HRI intern 2014

slide-2
SLIDE 2

Overview

  • Odometry

(noisy GPS / IMU for now)

  • Monocular camera
  • Lane level map
  • lateral localization on highway
  • give correct estimate on merge

/ split points Goal Input Assumptions

  • road surface is flat
slide-3
SLIDE 3

Coordinate System

North / East / Down Longitude / Latitude local plane origin = first gps position GPS Algorithm California State Plane Map

slide-4
SLIDE 4

Map

no semantic information, interpolate

slide-5
SLIDE 5

Particle Filter

  • Motion model

current pose previous pose GPS / odometry

  • Observation model

map current image pose = (north, east, yaw) map = a set of points labeled by marking groups

slide-6
SLIDE 6

Particle Filter

propagate using motion model weight each particle by its likelihood computed from observation model resample particles according to their weights all with same weight here

slide-7
SLIDE 7

Motion Model

initial rotation noise translation noise final rotation noise

  • rotations and translation

computed from odometry

slide-8
SLIDE 8

Observation Model

project map points to bird’s-eye view Given the vehicle pose,

  • ur bird’- eye view image is expected to look

like this ...

slide-9
SLIDE 9

Observation Model

project map points to bird’s-eye view Given the vehicle pose,

  • ur bird’- eye view image is expected to look

like this ...

slide-10
SLIDE 10

Observation Model

inverse perspective transform filter Hough line fitting … while what we really observe is ...

slide-11
SLIDE 11

Observation Model

expected

  • bserved

match lines

slide-12
SLIDE 12

Maximum Bipartite Matching

  • bserved

expected

slide-13
SLIDE 13

… not so simple

  • rder must be consistent

matches cannot be too far away some map lines may not be detected in the image # candidate matchings

slide-14
SLIDE 14

Likelihood Score

map line i detected line matched to map line i distance between the line pair

  • prob. of not

detecting a map line total number

  • f map lines
slide-15
SLIDE 15

Speed Up Matching share among particles

(0,0)(1,1)(2,2)(3,3)(4,4) (0,1)(2,3)(3,4) (0,0)(1,1)(2,2)(3,3)(4,4) (0,0)(2,1)(3,2)(4,3) (0,0)(1,1)(2,2)(3,3)(4,4)

  • Sample to obtain a very small set
  • f candidate matchings
  • For the rest of the particles,
  • nly evaluate these candidate

matchings

  • Exploit spatial correlation of

matchings among nearby particles

(map, detected) (0,0)(1,1)(2,2)(3,3)(4,4) (0,0)(2,1)(3,2)(4,3) (0,1)(2,3)(3,4)

  • Preferable to sample particles

spread out in space.

slide-16
SLIDE 16

Speed Up Matching search in previous map lines’ extent

map group 16 map: ipm g16: 0 g3: 1 g7: 2, 3 g8: 2, 3 g11: 3, 4

  • Keep track of extent of every

map line

  • For a new set of detected

lines, search matchings for each map line only within its extent

  • Exploit temporal invariance of

matchings for a single particle at different times

map group 3 map group 7 map group 8 map group 11

slide-17
SLIDE 17

Process Images

inverse perspective transform filter Hough line fitting

slide-18
SLIDE 18

Inverse Perspective Mapping

pitch yaw height measured by hand

slide-19
SLIDE 19

Inverse Perspective Mapping

slide-20
SLIDE 20

Inverse Perspective Mapping

slide-21
SLIDE 21

Inverse Perspective Mapping

slide-22
SLIDE 22

Top-hat Filter

high response if one side of an edge is very dark

slide-23
SLIDE 23

Top-hat Filter

* *

threshold threshold & high response if one side of an edge is very dark more robust for detecting dark-bright-dark patterns

slide-24
SLIDE 24

Steerable Filter Second derivative of Gaussian

separable rotated to arbitrary angle

slide-25
SLIDE 25

Map-guided Filtering

SC logic OR

...

slide-26
SLIDE 26

Steerable vs. Top-hat noisy image

top-hat map-guided steerable

slide-27
SLIDE 27

Take Advantage of Map

Motion model

  • more likely to go along the current lane
  • cannot move beyond road edges

Observation model

  • map-guided image filtering
  • map-guided line fitting
slide-28
SLIDE 28

Hough Transform Line Fitting

25 line segments detected by OpenCV’s probabilistic Hough transform 6 lines remains after merging

slide-29
SLIDE 29

Experiments on straight lanes

straight avg lateral error: 0.22, max: 1.35

slide-30
SLIDE 30

Straight lanes

slide-31
SLIDE 31

Deal with Curved Lanes

curved avg lateral error: 0.25, max: 0.98

slide-32
SLIDE 32

Deal with Curved Lanes

  • detect whether

the line is a curve (i.e. residual of a linear regression is large)

  • if so, match only

the bottom segment

slide-33
SLIDE 33

Steerable vs. Top-hat straight

steerable, avg lateral error 0.23, max 1.15 top-hat, avg lateral error 0.2, max .86

slide-34
SLIDE 34

Steerable vs. Top-hat curve

steerable, avg lateral error 0.3, max 0.79 avg lon. error 0.7, max 1.55 top-hat, avg lateral error 0.47, max 1.8 avg lon. error 0.67 max 2.83

slide-35
SLIDE 35

Effect of the Number of Particles

slide-36
SLIDE 36

Issues and Extensions

  • shadows
  • more general markings (urban environment)

○ stop-lines (longitudinal correction) ○ curved lanes ○ model-free

  • investigate how number of particle affects

performance