Transcript Skynet Quadrocopter

Skynet
An Autonomous Quatrocopter
Designed by
Andrew Malone
And
Bryan Absher
Introduction
•
•
•
•
•
Flying robot
Self stabilizing
Able to fly in preprogrammed patterns
Autonomous
Low cost
Outline
•
•
•
•
•
•
•
•
•
Block Diagram
PWM Control
Motor Driver Circuit
Wireless Communications
Sensors
Control System
Results
Applications
Future Improvements
Power Consumption
• Logic Power
– 2 x PICLF877A Microprocessor
• 0.6mA at 3 V and 4Mhz
– 3 x LY530ALH 1 Axis Gryroscope
• 5mA at 3 V
– ADXL335 3 axis Accelerometer
• 3 uA at 3 V
• Use 2 button batteries at 150mAh each
Motor Driving System
• Control of High-Current Motors with a
Microprocessor
• Microprocessor Output
– PIC16F877A
– 2V to 5V max ~25mA
• Motor
– GWS EDF50
– ~4 Amps at 10.8 V
PWM Characteristics
• Output Voltage is Simulated
– Device is Switched On and Off
• PIC PWM max 25mA
• Magnifies Motor Driving Concerns
– Inductance
– Generation
– Noise
– Power on Ground
System Requirements
• Extremely High Current Gain
– ~1000A/A
• 10V Maximum Output from 11V Supply
• High Current Output
– ~5A per Motor
• Fast Switching Time
– < 20µs
• Complete Electrical Isolation
– No Common Ground
Optical Isolation
• Anode and Cathode
Voltages drive infrared LED
• Light Modulates
Phototransistor Base
Voltage
• Complete electric isolation
• Cheap ($0.60 EE store)
• Fast (5 – 10 µs)
• TIL111
• Perfect for PWM
Darlington Transistor
TIP 122
– 5A Max Current
– β >1000 at 5A
– ~1V VCE
– < 20µs Switching Time
Delivery to Motor
• AC Output Interacts with Inductance
• Motors Prefer DC inputs
• Low-pass Filter
http://www.zen22142.zen.co.uk/Design/dcpsu.htm
Final Circuit Design
Wireless Communication
• IEEE 802.15.1 (Bluetooth)
–
–
–
–
Low power (100mA Tx, 20mA Rx)
Complex Protocol Stack
Small Network Size
Fast Data Rates (1.5 Mbit/s, or 3 Mbit/s)
• IEEE 802.15.4 Zigbee
–
–
–
–
Low power
Low overhead
Slower data rates
Large network size
Our Implementation
• Simple configuration
• UART communications
– 115 kBaud (Limited by PIC16LF877A)
– 3.3 V
• RN-41-SM
– Light weight
– Low power
– High data speed
• Good for tuning PID
Sensor Theory
• Accelerometer
– Charged cantilever
– Change in
acceleration changes
the capacitance of the
cantilever
Sensor Theory
• Gyroscopes
– MEMS gyroscopes consist of a vibrating
structure
– Angular velocity changes the vibration
Sensor Implementation
• Ideal implementation:
– Initial angles = arctan(x/z) and arctan(y/z)
– ω from gyroscope reading
– Subsequent angles = initial angles + ∫ω*dt
– Accelerations relative to ground derived from
accelerometer combined with gyroscope
angle readings
– Velocity = ∫a*dt
– Position = ∫v*dt
Sensor Implementation
• Accelerometers and Gyroscopes vary
widely from specification
– Accelerometer bias must be calibrated
– Gyroscope bias varies over time
• Inaccurate over long periods
• Readings can be corroborated using a
Kalman Filter
• Integrals rely on fast sampling rate
Sensor Implementation
Angle
– Assume gravity is greatest acceleration
– Angle = arctan(r/z)
– Extremely accurate
• Change in Altitude
– Integrate Z-axis acceleration
• Accurate for very small accelerations
System Control
• PID control
– Proven method
– Standard Tuning methods
• Ziegler–Nichols
– Effective at controlling high order systems
de
PID (e(t ))  Kp  e(t )  KI   e(t )  Kd 
dt
Our Control System
Results
•
•
•
•
PID-controlled power output
Accurate angular orientation measurement
Sufficient lift, battery life
Wireless feedback
Applications
• Aerial Displays
– MIT Flyfire
• Flying sensor network
• Autonomous
surveillance
Improvements
• 32 or 16bit ARM processor at 100 Mhz
• Horizontal motion measurements
– Local or Global GPS
– SONAR
• Environment sensors
– CO2
– Visual
– Wind Speed
• ZigBee mesh network
– Create a flying sensor network
– Distributed intelligence
• Kalman Filter
– Reduce noise in angle measurements