e-kiss-me/vins-with-imu-nn

vins mono with a IMU neural network

VINS-Mono-with-IMU-NN

About VINS-mono-with-IMU-NN

A VI SLAM system that integrates IMU neural network observations, developed based on vins-mono.

IMU neural network

IMU neural network using RNIN-VIO,the pretrained-model can be downloaded from:model(password:gwkw).

sample data

Because neural networks are suitable for handheld devices, we provide a data packet collected by iphone12,sample data can be downloaded from Rosbag(password ovci)

Build and Run

1 dependencise

1.1 ROS ROS Installation

    sudo apt-get install ros-YOUR_DISTRO-cv-bridge ros-YOUR_DISTRO-tf ros-YOUR_DISTRO-message-filters ros-YOUR_DISTRO-image-transport

we use libtorch1.8.2+cu102 for nural network inference.

1.2. Ceres Solver Follow Ceres Installation, remember to make install. (Our testing environment: Ubuntu 18.04, ROS Melodic, OpenCV 4.5, Eigen 3.3.3)

2 build

Clone the repository and catkin_make:

    catkin_make
    source ~/catkin_ws/devel/setup.bash

3 run

    roslaunch vins_estimator euroc.launch 
    roslaunch vins_estimator vins_rviz.launch
    rosbag play YOUR_PATH_TO_DATASET/MH_01_easy.bag 

Reference

[1]Chen D, Wang N, Xu R, et al. Rnin-vio: Robust neural inertial navigation aided visual-inertial odometry in challenging scenes[C]//2021 IEEE International Symposium on Mixed and Augmented Reality (ISMAR). IEEE, 2021: 275-283.

[2]Qin T, Li P, Shen S. Vins-mono: A robust and versatile monocular visual-inertial state estimator[J]. IEEE Transactions on Robotics, 2018, 34(4): 1004-1020.