Real-Time Dense 3D Mapping of Underwater Environments

Real-Time Dense 3D Mapping repository. This pipeline address the problem of real-time dense reconstruction for a resource constrained autonomous underwater robot.

Prerequisites

We have tested the library in 16.04 and 18.04, but it should be easy to compile in other platforms. A powerful computer (e.g. i7) will ensure real-time performance and provide more stable and accurate results.

C++11 or C++0x Compiler

We use the STL and chrono functionalities of C++11.

OpenCV

We use OpenCV to manipulate images and features. Download and install instructions can be found at: http://opencv.org. Tested with OpenCV 3.3.1 and OpenCV 4.5.5.

Eigen3

Required by g2o (see below). Download and install instructions can be found at: http://eigen.tuxfamily.org. Required at least 3.1.0.

Other dependencies

sudo apt install libceres-dev libgflags-dev

Building our pipeline and example

The code has been tested on Ubuntu 16.04 and Ubuntu 18.04, with gcc 7.5.

Clone the repository:

git clone https://github.com/ApdowJN/RTDense3DMapping.git

To build our pipeline:

cd IROS/
mkdir build/
cmake -DCMAKE_BUILD_TYPE=Debug ..
make -j $(nproc)

How To Run Example

Mexico Dataset

Download the Mexico dataset.
Extract images.zip to obtain the left/ and right/ image folders

The folder structure should mirror the following:

Mexico
-> left
    -> #######.png
    -> #######.png
-> right
    -> #######.png
    -> #######.png
-> colmap.ply
-> filename_mapping.csv
-> intrinsics.yaml
-> poses.txt

Execute the following command to run example in Mexico dataset

cd scipts
./run_mexico.sh

The path paramters inside the script may need to be changed to reflect where the data has been stored. To do so, open the script and modify the first two variables:

DATA_PATH=<path-to-dataset>
OUTPUT_PATH=<desired-output-path>

This example extends to all the other datasets.

Datasets

Here you can find links to additional datasets:

Data File Formats

Filename Mapping

The filename mapping must follow the format:

<timestamp>,<timestamp>.png
<timestamp>,<timestamp>.png
<timestamp>,<timestamp>.png
<timestamp>,<timestamp>.png
.
.
.

where the <timestamp> matches the timestamps in the poses.txt file.

Intrinsics

The intrinsics file must adhere to the following format:

%YAML:1.0

left: 
K: !!opencv-matrix
    rows: 3
    cols: 3
    dt: d
    data: [574.760004965, 0.0, 399.37765462, 0.0, 575.289618425,  288.60124274, 0.0, 0.0, 1.0]
D: !!opencv-matrix
    rows: 1
    cols: 4
    dt: d
    data: [0.0, 0.0, 0.0, 0.0]

right:
K: !!opencv-matrix
    rows: 3
    cols: 3
    dt: d
    data: [575.6337256, 0.0, 391.22448089, 0.0, 574.43588977, 286.86941794, 0.0, 0.0, 1.0]
D: !!opencv-matrix
    rows: 1
    cols: 4
    dt: d
    data: [0.0, 0.0, 0.0, 0.0]

# image resolution [ width x height ]
size: [800, 600]

# baseline (mm)
baseline: 142.20

# rotation matrix to as input to stereo rectify
R: !!opencv-matrix
    rows: 3
    cols: 3
    dt: d
    data: [0.9999531274148135, -0.0024947708286728633, 0.009355163912669094, 0.0022038626034325114, 0.99951762419007, 0.030978410595931,                    -0.009427935242973135, -0.030956341061845496, 0.9994762723472247]

# translation matrix to as input to stereo rectify
T: !!opencv-matrix
    rows: 1
    cols: 3
    dt: d
    data: [-0.1422002, 0.00018052, -0.00134952]

Poses

The poses file must adhere to the following format:

<timestamp> <tx> <ty> <tz> <qx> <qy> <qz> <qw>
<timestamp> <tx> <ty> <tz> <qx> <qy> <qz> <qw>
<timestamp> <tx> <ty> <tz> <qx> <qy> <qz> <qw>
<timestamp> <tx> <ty> <tz> <qx> <qy> <qz> <qw>
<timestamp> <tx> <ty> <tz> <qx> <qy> <qz> <qw>
.
.
.

If any of the formats change, the changes must be propagated to the data loading function in the src/util.cpp file.