IAB-RC Inverted Autonomous Balancer Remote Controlled April 18, 2008 Jude Collins Christopher Madsen
Final Presentation ➲ Technical aspects of the robot ➲ Prior art ➲ Schedule ➲ Finances
Overview ➲ Robot system overview—Chris Capabilities How it works Robot hardware ➲ Remote control—Jude Controller hardware PC N64 Controller PDA
Overview ➲ Literature/product search—Chris “Trajectory Tracking Control for Navigation of Self-contained Mobile Inverse Pendulum” (1994). Segway (2002). David P. Anderson's “nBot” (2003). ➲ Schedule and finances—Jude ➲ Questions
Robot overview
Capabilities ➲ Balance ➲ Stand up ➲ Lay down ➲ Slide ➲ Jump curbs ➲ Crash ➲ ?????
How does it work? ➲ Hardware calculates “lean” of the robot ➲ Wheels turn to prevent “lean” ➲ Forcing the robot to lean causes the robot to drive ➲ The robot learns to lean into external forces to keep balanced ➲ Spinning one wheel faster than the other causes the robot to turn ➲ Kicking the robot may make the robot angry
Robot system overview
Microcontroller ➲ MSP430F1611 ➲ Running 4 MHz ➲ 12 bit A/D with 8 pin- accessible inputs ➲ Two 16 bit timers ➲ v ➲ Programmed in C
Inertial Measurement Unit ➲ IDG-300 Gyroscope ➲ ADXL-330 Accelerometer
IDG-300 Gyroscope ➲ Dual-Axis rate gyroscope ➲ Operates by oscillating masses and capacitively measuring vibration caused by Coriolis effect. ➲ Sensitivity: 2 mV/deg/s ➲ Max rate: 500 deg/s ➲ Operating voltage: v
ADXL-330 Accelerometer ➲ Triple-axis accelerometer ➲ Micro-machined structure suspended over silicon by polysilicon springs. Plates mounted on moving structure and a fixed structure act as a variable capacitor in a filter circuit to measure acceleration. ➲ Sensitivity: ~300mV/g ➲ Max acceleration: ±3.0g ➲ Operating voltage: v (sensitivity is ratiometric)
PID Controller
Estimating pendulum orientation ➲ Integrating rate gyros is subject to drift errors. ➲ Accelerometers only work to determine orientation when not accelerating. ➲ Using both estimates together gives better estimate of orientation.
Bluetooth Radio ➲ Basically a breakout board for NXP's BGB203. ➲ Class 1 so has a 100m range ➲ 100 mW max transmitted power ➲ 3.3 volts ➲ 1 Mbps max UART
Remote Control
Computer ➲ Communicates via Bluetooth dongle ➲ Used for early verification of control law ➲ Jitter in transmission limited stability
The Remote Control ➲ Needed Peripherals Joystick A few buttons ➲ Modify old N64 controller. Exceeds requirements Cheap ($5-$15 on Amazon.com)
Control Flow
The microcontroller ➲ Needed peripherals UART High clock frequency Low Vcc ➲ ATMEGA8515L UART 20 MHz 2.7 – 5.5 V Low Cost ($3.06-$5.27 Digikey.com)
PDA ➲ Can also be used to remotely control robot. ➲ And ?????
Prior art ➲ First appearance of similar two-wheeled inverted pendulum that can navigate in 2 dimensions on a plane: “Trajectory Tracking Control for Navigation of Self-contained Mobile Inverse Pendulum” by Yunsu Ha and Shin'ichi Yuta of Japan in Position encoders on wheels (2000 step) Sensors to detect obstacles No remote-control
Prior art ➲ Segway Most popular inverted-pendulum type product. Patented just about everything imaginable concerning inverted pendulum human transportation. Have several Robotic Mobility Platform (RMP) models
Prior art ➲ David P. Anderson's nBot Received NASA's Cool Robot of the Week award and subsequently became well-known in the minds of robotics enthusiasts (2003). Launched a revolution of inverted-pendulum robot building. Homebrew shaft encoders.
How are we different? ➲ Back EMF encoders rather than mechanical encoders. ➲ Bluetooth radios enabling hardware-in-the- loop simulation. ➲ Stands up autonomously. ➲ Lays itself down gently. ➲ And ?????
Schedule
Finances—Robot
Finances—Remote Control
Finances ➲ Allotted budget: $1000 ➲ Expenditures: ~$500 ➲ Main expenses: 1 IMU -- $110 2 Bluetooth radios -- $120 1 MSP430 on breakout board -- $50 2 Motors and H-Bridges -- $80
Questions?