The autopilot CPU performs several functions in the node:
- using an inertial measurement unit (IMU) comprising 3 axis accelerometers and 3 axis gyroscopes provides stable flight for the airframe
- using a GPS+Compass combo it provides navigation to a desired waypoint (specified by absolute latitude and longitude)
- using a barometric altimeter it provides navigation to a desired altitude (relative to the take-off altitude)
- using an input from an FM receiver it allows for a mission abort
- provides a number of Failsafe modes not relying on external input
- the authoritative source of information for the Arducopter Autopilot
Installation and Calibration
Follow the setup and calibration steps in order:
- connection to airframe
- Select Frame type details
- motor setup (ensure that motors rotate in the correct directions)
- radio calibration
- calibrate the compass
- ESC Calibration details
- ESC calibration (original instructions)
- calibrate accelerometer
- calibrate battery sensor (ensure that the voltage pin is correctly set in the Mission Planner)
- To avoid excessive noise in the IMU readings, the APM should be mounted on vibration dampening foam; during initial hovering tests ensure that accelerometer noise is not higher than +/- 3 for X and Y axes and +/- 5 on the Z axis. This page explains how to measure the vibrations.
- Instructions for obtaining and installing the latest APM firmware?
DataLog Analysis
In case of problems with the drone, the datalogs on the autopilot may hold the answer to the problems encountered
Components
Good Choices
Our current choice of autopilot is the APM 2.6 (with external compass) from 3D Robotics Due to our choice of the electronics sandwich, the variant with side pin case (cables enter from side).
Sensors:
- Integrated IMU
- Integrated barometer (MS5611-01BA03) - only accurate to 1-2m - altitude readings appear to change with changes in air pressure (e.g., passing thermal)
- External GPS+Compass
- Battery Voltage and Current Sensor relatively inaccurate - needs work for calibration
Advantages
- Relatively inexpensive
- Open source firmware
- Uses MAVLink as it's communication protocol with the ground base station, allowing for integration with the sensor node?
- supports USB as well as wireless interfaces (e.g., XBee) for communication with the ground base station
- ease of installation as well as modular design
Disadvantages
- Relatively low power CPU, so the firmware cannot be used to do more work. Not a serious drawback in our case, as we have the sensor node? on board for any computing external to the piloting of the drone.
- The APM is geared toward hobbyists rather than researchers and therefore the documentation is not as detailed and complete as one would wish.
- The company (3D Robotics) seems to be transitioning to the next generation of autopilots (PX4).
Blind Alleys
The other autopilot we considered was based on the Paparazzi project at ENAC. We evaluated two platforms supporting paparazzi:
- Lisa/M 2.0 from Transition Robotics;
- Umarim Lite Bundle;
Although the CPUs of both Lisa/M and Umarim are more powerful than that of APM, we found the paparazzi based boards difficult to setup (both hardware and software), requiring considerably more time to put together, and less flexible in their use as they use a proprietary communication protocol with the ground base station.
