AP_VIO

This repository contains an implementation of an Equivariant Filter (EqF) for Visual Inertial Odometry (VIO), on the ArduPilot platform.

Required Dependencies

• C++17 (GCC Version >=10.1)

• CMake ( for build environment, Version >=3.12 ): sudo apt install cmake

• Eigen 3: sudo apt install libeigen3-dev

• Yaml-cpp: sudo apt install libyaml-cpp-dev

• GIFT ( for feature tracking ): https://github.com/pvangoor/GIFT

Required Hardware Setup

• Onboard computer (e.g Rapberry Pi)
• Global shutter USB camera (needs to be hard mounted)
• Autopilot instance with IMUs (e.g Cube Orange)

Building

To checkout and build AP_VIO on the Pi, execute the following at a shell:

git clone https://github.com/ShawnGYX/AP_VIO.git
cd AP_VIO
mkdir build
cd build
cmake ..
make 

Known Issue

• There is a known issue with data logging due to conflict between mutex locks in aofstream.h

Preparations

Camera Calibration

We use a checkerboard to calibrate the camera. Download the tools from https://github.com/STR-ANU/VIO-Tools.

Camera-IMU Calibration

We use Kalibr to calibrate camera to IMU. Available at: https://github.com/ethz-asl/kalibr. Can also be used for Camera calibration

Usage Instructions

Start VIO

To start VIO on the onboard computer, go to AP_VIO folder, execute the following command in the shell

./build/vio_ap <config_file> <serial port>

for example: ./build/vio_ap EQVIO_config_simple.yaml /dev/serial0.