QUADCOPTER Group 1

• Floris Driessen • Martin van Leussen • Art Senders • Tijl van Vliet • Steven van der Vlugt

Huginn

Presentation outlines • Assignments and goals

• A) Build the quadcopter and get it flying

• B) PID tuning

• C) Rotate 360 degrees around the z-axis

• D) Marker detection

• E) Environment map

• Results

• Recommendations

A) Hardware setup

Base Station

Quad Copter 4x motor

PC Manual

GUI 9Ch

Transmitter

MultiWii Pro IMU RF Bluetooth

ESC

A) Multiwii

• Active community • Open source • Good quality • Powerful controller

Accelerometer

Gyroscoop Magnetometer

Atmega2560

B) Stable Quadcopter • Flash ESC with custom firmware

• PID tuning

• Ziegler Nichols

• Impulse response

• Step response

B) Stable Quadcopter

-20

-10

0

10

20

30

40

50

60

70

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97 100

103

106

109

112

115

118

121

124

127

130

133

136

Pitc

h in

tent

h of

deg

rees

Time

B) Stable Quadcopter

-90

-70

-50

-30

-10

10

30

50

70

90

110

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

106

111

116

121

126

131

136

141

146

151

156

161

166

171

176

181

186

191

196

201

206

211

216

221

226

231

236

241

246

251

256

Pitc

h in

tent

h of

deg

ree

Time

Pitch

C) Rotate 360 degrees around the z-axis • Panorama • Location hold

• Fixed height • Fixed location

C) Hardware setup

Base Station

Quad Copter

Beagle Board

4x motor

Bottom Cam Kinect

PC Manual

Console

Point-cloud

Top Sonar

Bottom Sonar

9Ch Transmitter

MultiWii Pro IMU RF

WiFi

Sonar Controller ESC

C) Height control • Fix altitude

• Ultrasonic sensor

• Pitch correction

Measured Height

Height Correction

C) Height measurement

-400

-300

-200

-100

0

100

200

300

400

60

65

70

75

80

85

90

95

100

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79

Angl

e in

tent

hs o

f deg

rees

Hei

ght (

cm)

Time

Measured altitude

Corrected altitude

pitch

C) Control • Start quad copter with RF transmitter • Enable altitude hold • Switch roll, pitch and yaw control to BeagleBoard

Pitch Roll Throttle

Yaw

C) Global control (HW)

C) Global control (SW) • Ubuntu

• Cross platform • Experience and know how

• OpenCV • Powerful open source computer vision library by Intel • General purpose vision functions • Functions to work with video streams

• Open Kinect • Light weight open source library

C) Fixed location • Optical flow

• Stabilization based on detected features

• Marker • Stabilization based on marker orientation

C) Optical Flow

C) Optical Flow 50 strongest features

outlier 350 strongest features

C) Fixed location • Optical flow

• Requires trajectory planning to navigate the quadcopter

• Marker • Marker can be followed throughout a building

D) Marker detection • Follow line instead need for trajectory planning • Take a corner in a line

• Line detection

• Houghline: vector extraction from edges • Classification: combine vectors to lines

• Marker identification

D) Image processing 1. Color image 2. Blur (to remove noise from

processing) 3. Black and white image

(not conventional) 4. Canny edge detection 5. Houghlines transform

(vector output) 6. Vectors to lines (classification) 7. Detect crossing + type

D) Marker classification algorithm • Map 8 fields next to the crossing of the lines • 8 locations to identify different markers • Corner detection • Cross detection for 360° rotation • Line following

D) Marker classification algorithm

Low fps due to manual debugging

Bitfields of squared where a line is detected

0

1

Angles of lines detected

D) Control loop

D) Global control (SW)

E) Environment mapping • Virtual 3d map of the environment • Navigation purposes (Autonomous navigation) • Search and rescue • Digitalization of buildings

E) Point cloud • Kinect Depth image

• UDP Stream with depth image from Beagle Board

• Use PCL (Piont Cloud Library) for stitching

• Mixture Java and C++

• JNI for java to C++ communication

E) Point cloud program

E) Stitching phases

Existing

New

Alignment Difference New frame

Update existing

E) Iterative Closest Point steps 1. Find for each point the closest point in the second

cloud. 1. Linear search 2. K-nearest neighbors search

2. Find the transformations per previews found points. 1. Mean square cost function

3. Transform the points. 4. Iterate this process.

Results • A) Quadcopter is built

• B) Is flying stable

• C+D) Algorithm are working correctly

• But cannot track the line over time

• Too small webcam angle

• Too slow processing

• E) Showing mapping

• Mapping does not grow

• ICP not working correctly

Quadcopter result

Recommendations • Wide angle lens • Acceleration of algorithms

• DSP • Resolution down scaling

• Mapping • Do not use JNI within mapping program • Build program complete in PCL

Demo

Questions?

? ? ?

Obstacle Detection

Global control (HW)

Became available after start of the project

E) Iterative Closest Point Phases