The following discussion concerning accelerometer samples is divided into four scenarios: 1. Low-speed on smooth terrain, 2. High-speed on smooth terrain, 3. Low-speed on rough terrain, and 4. High-speed on rough terrain. Each of those scenarios present raw accelerometer data and filtered accelerometer data using a non-weighted average of 5 samples.

All scenarios utilized a MicroStrain 3DM orientation module configured to output raw accelerometer and magnetometer data (for the purposes of this analysis the magnetometer data was disregarded.) The 3DM module was mounted in the rear of the Urban Robot, directly along the centerline. All accelerometer data reported is G units, translated from the raw accelerometer data using the following MicroStrain provided conversions:

Force_in_G’s = 2 * (Accel_Data – ACCEL_OFFSET- 2048) / ACCEL_GAIN

Accel_Data = 16 bit value returned from module (one 16 bit value per axis)

ACCEL_OFFSET = Factory set value stored in 3DM module EEPROM (one 16 bit value per axis)

ACCEL_GAIB = Factory set value stored in 3DM module EEPROM (one 16 bit value per axis)

**NOTE: It is easiest to obtain the offset and gain values using the MicroStrain provided interface software. However, it is also possible to directly read (and write) from the module’s EEPROM. In this case, to obtain the ACCEL_OFFSET value used above, the value returned from EEPROM must be added to –32768. That is:

ACCEL_OFFSET=ACCEL_OFFSET – 32768

For all scenarios presented, the accelerometer sampling rate is 40Hz. For all scenarios presented, data sampling began and ended with the robot in stationary position.

It is apparent from data collected on a reasonably level and stationary robot, that there exist significant inaccuracies in the device. The magnitude of the force experienced by the module in this situation calculates to only .95G’s.