Welcome to the Hand-Held Mobile Mapping System for Large-Scale Surveys workshop. During the workshop attendees are expected to perform mapping tasks with handheld devices and process results. The device is an open hardware project available here: mandeye.

You can assemble your device, but we will bring a dozen devices for you.

Notes 💡

NOTE1: Please revisit this page before the workshop to get the latest updates

NOTE2: All refered gitthub projects mandeye, HDMapping are open to you contributions

Prerequesities

The project HDMapping is supported for Windows and Linux. We recommend the following distributions:

  • Windows 11
  • Windows 10
  • Ubuntu 22.04
  • Ubuntu 24.04

The machine you are running should have plenty of RAM. For small-scale experiments, 16 GB should be enough. Consider adjusting virtual memory to utilize disk space, details can be found here

Extra software

It is not essential, but highly recommended to download install CloudCompare:

sudo apt-get install CloudCompare

Software

The project is available with binaries both for Windows and Linux.

Ubuntu, prebuilt packages

The prebuilt package is done automatically and can be found here for version 0.61.0 0.62.0

sudo apt-get install freeglut3-dev libeigen3-dev liblaszip-dev nautilus
sudo dpkg -i hd_mapping-0.61.0-Linux.deb

Windows, prebuilt packages

Visit release page and download binaries.

Build from source

The HDMapping project, can be built from source. It comes with all necessary 3rd party libraries.

git clone https://github.com/MapsHD/HDMapping.git
cd HDMapping
mkdir build
git submodule init
git submodule update --recursive
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make -j

Sample Dataset to download

We will work with the data you capture, however please download following datasets - as more advanced usecase.

Underground parking

Ground truth:
Dataset:

Download link:
link

Roccastrada

Dataset:
Download link:
link

Forest

Download link:
link

Rubble Field

Download link:
link

Cave

Download link:
link

3rd party SLAM evaluation with HDMapping

We prepared two 3rd party SLAMs to evaluate:

  • Kiss ICP
  • FAST Lio

Those are run docker environment. Please checkout repository:

git clone https://github.com/michalpelka/Slams_dojo.git
cd Slams_dojo
git submodule init
git submodule update

Next run chose SLAM with some demo dataset. Refer to readme for more details. Here is an example to run FAST LIO with osrf/rocker :

docker build ./slams/fastlio/ -t dojo_fastlio_parking --build-arg DATASET=data_undeground_parking 
rocker --x11 --nvidia auto --volume $(pwd)/evaluation-fastlio:/evaluation -- dojo_fastlio_parking ./system_run.sh

Here is expected results:

NOTE3: Please build those images to have better experience on workshop:

cd Slams_dojo
docker build ./slams/fastlio/ -t dojo_fastlio_usersdata --build-arg DATASET=users_data
docker build ./slams/kiss-icp/ -t dojo_kissicp_usersdata --build-arg DATASET=users_data

Recent Posts