/pyslam

pySLAM contains a monocular Visual Odometry (VO) pipeline in Python. It supports many modern local features based on Deep Learning.

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

pySLAM v2

Author: Luigi Freda

pySLAM contains a python implementation of a monocular Visual Odometry (VO) pipeline. It supports many classical and modern local features, and it offers a convenient interface for them. Moreover, it collects other common and useful VO and SLAM tools.

I released pySLAM v1 for educational purposes, for a computer vision class I taught. I started developing it for fun as a python programming exercise, during my free time, taking inspiration from some repos available on the web.

Main Scripts:

  • main_vo.py combines the simplest VO ingredients without performing any image point triangulation or windowed bundle adjustment. At each step $k$, main_vo.py estimates the current camera pose $C_k$ with respect to the previous one $C_{k-1}$. The inter-frame pose estimation returns $[R_{k-1,k},t_{k-1,k}]$ with $||t_{k-1,k}||=1$. With this very basic approach, you need to use a ground truth in order to recover a correct inter-frame scale $s$ and estimate a valid trajectory by composing $C_k = C_{k-1} * [R_{k-1,k}, s t_{k-1,k}]$. This script is a first start to understand the basics of inter-frame feature tracking and camera pose estimation.

  • main_slam.py adds feature tracking along multiple frames, point triangulation, keyframe management and bundle adjustment in order to estimate the camera trajectory up-to-scale and build a map. It's still a VO pipeline but it shows some basic blocks which are necessary to develop a real visual SLAM pipeline.

You can use this framework as a baseline to play with local features, VO techniques and create your own (proof of concept) VO/SLAM pipeline in python. When you test it, consider that's a work in progress, a development framework written in Python, without any pretence of having state-of-the-art localization accuracy or real-time performances.

Enjoy it!

VO

SLAM

Feature Matching


Install

Clone this repo and its modules by running

$ git clone --recursive https://github.com/luigifreda/pyslam.git

The framework has been developed and tested under Ubuntu 18.04. Use the available specific install procedure according to your OS:

  • Ubuntu 18.04 =>
  • Ubuntu 20.04 and Ubuntu 22.04 =>
  • MacOs =>
  • Windows =>
  • Docker =>

Requirements

  • Python 3.6.9
  • Numpy (1.18.2)
  • OpenCV (4.5.1 and newer versions supported, see below for a suggested python installation)
  • PyTorch (>= 1.4.0)
  • Tensorflow-gpu 1.14.0

If you run into troubles or performance issues, check this TROUBLESHOOTING file.


Ubuntu 18.04

Install in your working Python environment:

If you want to launch main_vo.py, run the script:

$ ./install_basic.sh

in order to automatically install the basic required system and python3 packages.

If you want to run main_slam.py, you must additionally install the libs pangolin, g2opy, etc. by running:

$ ./install_all.sh

Install in a custom Python virtual environment:

If you do not want to mess up your working (base) python environment, you can create a new virtual environment pyslam with venv by easily launching the scripts described here.

If you prefer conda, run the scripts described in this other file.

N.B.: a single python environment is able to support all the supported local features!


Ubuntu 20.04 and Ubuntu 22.04

This procedure is valid for both Ubuntu 20.04 and Ubuntu 22.04. Clone this repo recursively and move into the branch ubuntu20

$ git clone --recursive https://github.com/luigifreda/pyslam.git
$ cd pyslam 
$ git checkout ubuntu20  

and then follow the instructions for creating a new virtual environment pyslam described here.


MacOS

Check the instructions in this file.


Docker

If you prefer docker or you have an OS that is not supported yet, you can use rosdocker with its custom pyslam or pyslam_cuda docker file.


How to install non-free OpenCV modules

The install scripts take care of installing the new available opencv version (4.8.1 on Ubuntu 18) and its non-free modules.

Check your installed OpenCV version:
$ python3 -c "import cv2; print(cv2.__version__)"

Check if you have non-free OpenCV module support (no errors imply success):
$ python3 -c "import cv2; detector = cv2.xfeatures2d.SURF_create()"

For an advanced OpenCV installation procedure, you can take a look here.

Troubleshooting

If you run into issues or errors during the installation process or at run-time, please, check theTROUBLESHOOTING.md file.


Usage

Once you have run the script install_basic.sh, you can immediately run:

$ python3 -O main_vo.py

This will process a KITTI video (available in the folder videos) by using its corresponding camera calibration file (available in the folder settings), and its groundtruth (available in the same videos folder). You can stop main_vo.py by focusing on the Trajectory window and pressing the key 'Q'.

N.B.: as explained above, the basic script main_vo.py strictly requires a ground truth.

In order to process a different dataset, you need to set the file config.ini:

  • select your dataset type in the section [DATASET] (see the section Datasets below for further details)
  • the camera settings file accordingly (see the section Camera Settings below)
  • the groudtruth file accordingly (ee the section Datasets below and check the files ground_truth.py and convert_groundtruth.py )

Once you have run the script install_all.sh (or your specific OS install script as detailed above), you can test main_slam.py by running:

$ python3 -O main_slam.py

This will process a KITTI video (available in the folder videos) by using its corresponding camera calibration file (available in the folder settings). You can stop it by focusing on the opened Figure 1 window and pressing the key 'Q'.

You can choose any detector/descriptor among ORB, SIFT, SURF, BRISK, AKAZE, SuperPoint, etc. (see the section Supported Local Features below for further information).

Some basic test/example files are available in the subfolder test. In particular, as for feature detection/description/matching, you can start by taking a look at test/cv/test_feature_manager.py and test/cv/test_feature_matching.py.

N.B.:: due to information loss in video compression, main_slam.py tracking may peform worse with the available KITTI videos than with the original KITTI image sequences. The available videos are intended to be used for a first quick test. Please, download and use the original KITTI image sequences as explained below.


Supported Local Features

At present time, the following feature detectors are supported:

The following feature descriptors are supported:

You can find further information in the file feature_types.py. Some of the local features consist of a joint detector-descriptor. You can start playing with the supported local features by taking a look at test/cv/test_feature_manager.py and test/cv/test_feature_matching.py.

In both the scripts main_vo.py and main_slam.py, you can create your favourite detector-descritor configuration and feed it to the function feature_tracker_factory(). Some ready-to-use configurations are already available in the file feature_tracker.configs.py

The function feature_tracker_factory() can be found in the file feature_tracker.py. Take a look at the file feature_manager.py for further details.

N.B.: you just need a single python environment to be able to work with all the supported local features!


Datasets

You can use 4 different types of datasets:

Dataset type in config.ini
KITTI odometry data set (grayscale, 22 GB) type=KITTI_DATASET
TUM dataset type=TUM_DATASET
video file type=VIDEO_DATASET
folder of images type=FOLDER_DATASET

KITTI Datasets

pySLAM code expects the following structure in the specified KITTI path folder (specified in the section [KITTI_DATASET] of the file config.ini). :

├── sequences
    ├── 00
    ...
    ├── 21
├── poses
    ├── 00.txt
        ...
    ├── 10.txt

  1. Download the dataset (grayscale images) from http://www.cvlibs.net/datasets/kitti/eval_odometry.php and prepare the KITTI folder as specified above

  2. Select the corresponding calibration settings file (parameter [KITTI_DATASET][cam_settings] in the file config.ini)

TUM Datasets

pySLAM code expects a file associations.txt in each TUM dataset folder (specified in the section [TUM_DATASET] of the file config.ini).

  1. Download a sequence from http://vision.in.tum.de/data/datasets/rgbd-dataset/download and uncompress it.

  2. Associate RGB images and depth images using the python script associate.py. You can generate your associations.txt file by executing:

$ python associate.py PATH_TO_SEQUENCE/rgb.txt PATH_TO_SEQUENCE/depth.txt > associations.txt
  1. Select the corresponding calibration settings file (parameter [TUM_DATASET][cam_settings] in the file config.ini)

Camera Settings

The folder settings contains the camera settings files which can be used for testing the code. These are the same used in the framework ORBSLAM2. You can easily modify one of those files for creating your own new calibration file (for your new datasets).

In order to calibrate your camera, you can use the scripts in the folder calibration. In particular:

  1. use the script grab_chessboard_images.py to collect a sequence of images where the chessboard can be detected (set the chessboard size therein, you can use the calibration pattern calib_pattern.pdf in the same folder)
  2. use the script calibrate.py to process the collected images and compute the calibration parameters (set the chessboard size therein)

For further information about the calibration process, you may want to have a look here.

If you want to use your camera, you have to:

  • calibrate it and configure WEBCAM.yaml accordingly
  • record a video (for instance, by using save_video.py in the folder calibration)
  • configure the [VIDEO_DATASET] section of config.ini in order to point to your video.

Contributing to pySLAM

I would be very grateful if you would contribute to the code base by reporting bugs, leaving comments and proposing new features through issues and pull requests. Please feel free to get in touch at luigifreda(at)gmail[dot]com. Thank you!


References

Suggested books:

Suggested material:

Moreover, you may want to have a look at the OpenCV guide or tutorials.


Credits


TODOs

Many improvements and additional features are currently under development:

  • loop closure
  • relocalization
  • map saving/loading
  • modern DL matching algorithms
  • object detection and semantic segmentation
  • 3D dense reconstruction