quadcopter navigation using aakash tablet with on board image processing
DESCRIPTION
Final Presentation for Team Garuda . This quadcopter and its ground control station(GCS) was built in the Integrated Development Lab during my summer internship at IIT Bombay . https://www.youtube.com/watch?v=0966YNqcisg.TRANSCRIPT
QUADCOPTER NAVIGATION USINGAAKASH TABLET WITH ON BOARD IMAGE
PROCESSING
Garuda Control Station
Principal InvestigatorProf. D.B. Phatak
TEAM GARUDA
Project Team Members
Mr. Aaditya Hambarde
Mr. Ankit Kumar
Mr. D Yogendra Rao
Mr. Shubham Garg
Miss. Shuchi Sharma
Project Mentors
Mr. Sandesh PatiMiss Maitreyee MordekarMr. Sivakiran KodaliMr. Varun Madkaikar
Project In-Charge Mr. Rajesh Kushalkar
FEATURES
• Video Streaming
• Door detection
• GPS Tracking
• Ground Control Station on Aakash
Project Overview
YAW
ROLL
PITCH
YAW CONTROL
ROLL/PITCH CONTROL
‘IF YOU WANT TO GO FORWARD, PITCH FORWARD’
ALTITUDE CONTROL
PROPELLERS
• Two clockwise and two anti-clockwise
• Need a good balance on deciding length and pitch
Courtesy : www.atomikrc.com
ELECTRONICS SPEED CONTROLLER
• Generates signals for Brushless DC motor
• Pulse duration for 1000 us – 2000us
Courtesy : www.hangarone.co.nz
ACCELEROMETER & GYROSCOPE (MPU6050)
• Accelerometer :Sensitivity to vibrations
• Gyroscope : drift
Angular rate in three axes of space in deg/s
Courtesy : https://community.freescale.com/community/the-embedded
MAGNETOMETER (HMC5883L)
• Digital-output : angular movement , magnetic field intensity
• Reliable direction for magnetic north
Courtesy : https://community.freescale.com/community/the-embedded
PRESSURE SENSOR (MS5611)
• Altitude of quadcopter
• Temperature of the environment
Pressure is our 10th axis
Courtesy : https://community.freescale.com/community/the-embedded
GPS SENSOR (MT3329)
• Triangulation Method
• Location of quadcopter
• NMEA Protocol
(National Marine Electronics Association)
Courtesy : http://www.trimble.com/gps_tutorial/howgps-triangulating.aspx
SENSOR FUSION
• It literally means ‘fusing the data of two sensors’
• Accelerometer is noisy (has spikes).
• Gyroscope output has a time varying drift.
• Sensor Fusion-We take the best of both the worlds.
EXTERNAL AUTONOMOUS NAVIGATION
• Various Flight Modes-Stabilize, Acro, Alt Hold, Loiter, Auto Mode.
• For the AUTO mode, GPS is necessary.
• Waypoints can be set by our Application on the AAKASH tablet.
INTERNAL AUTONOMOUS NAVIGATION
• GPS fails in a closed-door environment.
• Detect a door/window and go out where GPS access is present.
• In future, we hope to design a controller for navigation of quadcopter from indoor to outdoor environs.
• SONARs will be used to create a depth map of the indoor environment and for obstacle detection.
Courtesy: http://www.scientificamerican.com/
DOOR DETECTION ALGORITHMS• Various alternatives like Training by Haar cascade, Feature matching by SURF
features were explored.
• To get a reasonable FPS, we had to settle on an algorithm which exploits the geometric properties of the door.
PID TUNINGProportional(P)Integral(I)Derivative(D)
1. NEED FOR PID TUNING
• For a stable and smooth flight of the quadcopter• To get the optimum or desired behavior from it•Without this, quad may topple, become sluggish or untamable
COMMUNICATION PROTOCOL : MAVLINK
• It is a protocol for communication between a Ground Control Station (GCS) and unmanned vehicles
• Each MavLink packet has a length of 17 bytes
6 bytes header + 9 bytes payload + 2 bytes checksum
RC INTERFACE
Fast Mode Access
Navigation Menu
RC Control
Connect Button to connect with
the quad
Mode Switch
INTERFACES
Planning Interface
Flight Data Interface
INTERFACES
Charts Interface
Camera Feeds
WHAT DID WE LEARN?
• Orientation algorithms of a quadcopter.
• Aerodynamics of a quadcopter.
• Explored Image Processing Techniques in detail.
• Streaming the video-feed of a USB webcam/RaspiCam over a local area network through MJPEG streamer.
• A new way of thinking and questioning things.
• A new communication protocol-MAVLINK.
CHALLENGES FACED
• Implementation of Haar training for door detection.
• PID tuning for stable flight.
• Overlapping the RC on the video feed.
FURTHER PROSPECTS
• Making ‘Garuda’ autonomous (using GPS based navigation).
• Designing a PID controller for internal navigation by door/window detection and SONAR.
• Face Recognition in the feed received at the AAKASH end.
• Recording the video feed on AAKASH tablet and processing it.
REFERENCES
• [1] https://github.com/AngusP/SkiTrack/blob/master/documentation/MS5611-01BA03%20datasheet.pdf
• [2] http://www.invensense.com/mems/gyro/.../PS-MPU-6000A-00v3.4.pdf
• [3] http://www.adafruit.com/datasheets/HMC5883L_3-Axis_Digital_Compass_IC
• [4] JoyStickView : https://code.google.com/p/mobile-anarchy-widgets/
• [5] MjpegStream : https://bitbucket.org/neuralassembly/simplewebcam
REFERENCES
• [6] HUD : https://code.google.com/p/copter-gcs
• [7] MAVlink : Mavlink for Dummies Part 1
• [8] Barton, Jeffrey D. "Fundamentals of small unmanned aircraft flight." Johns
• Hopkins APL technical digest 31, no. 2 (2012): 132-149.
• [9] Premerlani, William, and Paul Bizard. "Direction cosine matrix imu: Theory."DIY
• Drones.[Online][Cited: 1 7 2012.] http://diydrones. ning. com/profiles/blogs/dcmimu-theory-first-draft (2009).
• [10] http://dev.ardupilot.com/
THANK YOU