...
| Description | Tracks accelerometer and gyroscope data on CAN Bus | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Target release | |||||||||||
| Epic |
| ||||||||||
| Document owner | |||||||||||
| Project leads | |||||||||||
| Team members |
Goals
- Relays raw accelerometer and gyroscope data to controller board on CAN Bus
Grades and performance categories
The grade and performance list is organized from high-end expensive IMU systems that provide Inertial Navigation Systems to low-end accelerometer and gyroscope configurations:
- Marine and Navigation
- Tactical
- Industrial
- Automotive & Consumer
There are two grades applicable to our need: Automotive, and Industrial.
Automotive vs Industrial
Automotive grade sensors are typically sold as separate accelerometers and gryoscopes, so it is possible to combine many accelerometers and gyroscopes from different manufactures into one IMU. The main difference between an Automotive grade and Industrial grade sensor is in the sensor calibration. Without good calibration, the automotive grade sensor will create large errors in angular velocity rates and orientation data. Automotive grade sensors are not good for position tracking even with the aid of a GPS unit, due to its position drift errors. With good calibration it can be useful for angular velocity, and pitch/yaw/roll measurements.
IMU Measurement errors
The errors specific to angular velocity and orientation measurements are listed below:
Dynamic errors due to gyro scale factor
This is the main concern for using the automotive grade sensor in tracking the orientation of the car. The gyro scale factor correctly scales the measurement during the motion, and if the scale factor is off significant errors after the motion will be present. The automotive grade sensor is useless in inertial navigation systems and in tracking orientation without rigiourous and precise calibration. With proper calibration, it can perform basic navigation techniques.
Dynamic errors due to gyro misalignment
This is the main concern for using an automotive grade sensor in tracking angular velocity. Any gryo misalignment will skew the angular velocity in all coordinate directions. Rotating the sensor in one axis for 360 degrees and placing it down will result in angular rates being recorded in the all other axis that are misaligned, even when there was no change in that axis.
Static errors due to accelerometer bias
The errors from accelerometer bias arise in automotive grade sensors due to imprecise calibration. The bias changes the measured acceleration vector to an incorrect direction, and after subracting the gravitational acceleration the resulting acceleration will result in errors in the pitch and roll.
For industrial grade sensors, these errors are due to temperature changes.
Static errors due to accelerometer misalignment
These are removed after a tumble test during the calibration procedure.The magnitude of error is of no concern here.
Static errors due to accelerometer scale factor
The error in orientation is due to remaning quantities after the calibration process. A scale factor error in one axis will stretch a vector in that dimension, affecting the measured angle.
Background and strategic fit
The sensor board is used to obtain data and diagnostics of the motion of the car (i.e. pitch, yaw, roll). Additional features of the board are to be decided, based on what other metrics or data, such as humidity or temperature, is useful to have in the car. The sensors board will be connected to the main CAN Bus.
...
- Log acceleration and angular velocity of the car
- Support timestamped data so it can be correlated with GPS, power data, pedal, steering, etc.
- Locate the sensor as close to the center of gravity of the car as possible
Background and strategic fit
Mechanical has requested an IMU to correlate their models for vehicle dynamics with real data. This sensor should be located near the center of gravity of the car. Specific requirements are listed in the main ticket.
Additional features of the board such as humidity or temperature, are yet to be decided.
Assumptions
- The data does not need to be broadcast in realtime
- The data can be processed offline
- We do not need to sample very quickly (~5Hz)
Requirements
# | Title | User Story | Importance | Notes |
|---|---|---|---|---|
| 1 | Records motion data |
References
Vectornav.com. (2017). IMU and INS - VectorNav Library. [online] Available at: https://www.vectornav.com/support/library/imu-and-ins [Accessed 30 Sep. 2017].| Mechanical would like to verify their vehicle dynamics models against our motion data. | Must Have | Mostly accelerometer, gyroscope, magnetometer for angular velocity and pitch/yaw/roll. Data must be timestamped. | ||
| 2 | Logging | We need to be able to store the data somewhere so it can be processed. | Must Have | We should consider how to log the data without flooding the CAN bus or XBee. |
| 3 | Post-processing | It would probably be nice to be able to visualize and use the data that we've collected. | Nice to have | This may be offloaded to Mechanical. |
| 4 | Multiple sensor boards | It might be nice to support multiple sensor boards so we can get a better idea of how our car works. | Nice to have |