Robot Hardware and Control

1
Robot Hardware and Control

• Sarah Bergbreiter
• UC Berkeley
• June 17, 2002

2
Why Use TinyOS for Robots?

• Hardware
• Hardware Modularity through Data Bus
• new boards are simple to add and interface
• Wireless Connectivity through RFM Radio
• Cheap and off-the-shelf
• Software
• Event driven operating system convenient for
  behavior based control
• Hardware abstraction provides a more simple
  interface for the application developer

3
Robot Hardware

• 2-channel MotorBoard provides an interface
  between Rene I/II motes and robot platform
• Currently using an off-the-shelf RC car (Kyosho
  Mini-Z)

• Each H-bridge allows variable speed and direction
  to the motors through a PWM signal sent to the
  motors
• Whisker Board is also available (binary contact)

4
New Robot Hardware

• New Robot MotorBoard is currently being tested to
  interface with MICA and future NEST motes
• Includes an Atmel ATmega8L processor
• handles communication protocols
• sends PWM signals to motors
• accelerometer on board for odometry
• New motor driver ICs from Intersil

5
The Software Interface

• Three Simple Commands
• setspeed(0-255)
• setdir(FORWARD, REVERSE)
• turn(0-40), 20 STRAIGHT
• Hardware abstraction hides
• changes in the hardware from
• application developer

6
Where it all comes together...
Autonomous, Inexpensive, Wireless,
Easy-to-Program, Modular Robots!
7
Open Loop Control

• Open Loop Control involves a robot following a
  programmed path without any feedback as to where
  it is on that path
• Path Programmed at
• Install time (can not be changed)
• Over wireless Paths may be updated or generated
  on the fly

8
Can we close the loop through the Network?

• No positioning sensors currently on the robots
• Would like to obtain position information from
  the network
• Using this estimated position, we can close the
  loop

9
Closed Loop Control

• Simulated Position Estimation Data provided by
  mouse clicks
• PID control loop

10
Closed Loop Control
11
Where to go from here

• Integrate with real position estimates from over
  the network
• Integrate with System ID information (Bruno
  Sinopoli) and use state estimates to improve
  position information
• Demonstrate closed loop control from the network

12
Thank you.