B-B-B-Balancing

So first off, a Segway needs to know at what angle it is to the ground. This can be accomplished surprisingly easily.


Their are basically two types of sensor used for sensing tilt in a robotics application, the Gyro and the Accelerometer. Both have their advantages and disadvantages, and both must be used for reliable tilt data.

Accelerometer - Used in the Nintendo Wiimote and Nunchuck

Pros

• Doesn't Drift
• Outputs Acceleration due to gravity

Cons

• Very prone to vibration
• Not very precise
• Also senses the acceleration of the Scooter

Gyro - Used in Helicopters

Pros

• Very precise
• Very smooth input
• Provides angular rate
  

Cons

• Drifts - A LOT!

So what do you do?

Combine these two inputs so that you can get a solid, quickly responding signal, that doesn't drift. There are two common ways to accomplish this:

1. Kalman Filter
The Kalman filter is a mathematically complex filter that provides a theoretically ideal combination and smoothing of sensor signals.

Implemented on the Arduino Here

ArduIMU

Might be tried, but not until a working system is created

2. Angle Complementary Filter
Uses a time bias to determine which sensor to trust
Described and implemented in the SegSpecs.zip file here
Easier to implement and cancels out drift and vibration.

Comparision: Complementary Filters vs Kalman Filter vs Extended Kalman Filter