Why MOLA¶
MOLA is a production-grade LiDAR SLAM framework for robotics, autonomous vehicles, and 3D surveying. It follows an Open Core model: the full framework is open-source and free for research, while commercial users can upgrade to MOLA Pro for closed-source deployment, pre-built binaries, and priority support.
What makes MOLA different¶
Fully configurable - no recompilation¶
MOLA pipelines are defined entirely in YAML. Swap ICP matchers, change map layers, adjust filter chains, or switch between LO and LIO - all without rebuilding a single line of C++. This means faster iteration during development and safer updates in production.
Map-less geo-referenced localization¶
Using the smoother state estimator to fuse LiDAR + IMU + GNSS (and wheel odometry, if available), MOLA estimates the vehicle pose in UTM/ENU and latitude/longitude without a prebuilt map. The trajectory keeps the centimeter-level local accuracy of LiDAR-inertial odometry, and its absolute accuracy is that of the GNSS: centimeter-level with an RTK receiver, and meter-level with a consumer-grade receiver, refined by fusing many fixes along the trajectory. This is ideal for outdoor robots, agricultural vehicles, and autonomous driving applications where absolute positioning matters.
Rich metric map ecosystem¶
MOLA’s .mm metric map format comes with a full set of CLI tools:
mm-viewer: Interactive 3D map viewer
mm-render: Render map layers into images, off-screen (no GUI)
mm2las, mm2ply, mm2txt: Export to LAS, PLY, TXT for GIS/surveying workflows
mm2kml: Export georeferenced maps and trajectories to Google Earth
mm-filter: Filter and transform maps
mm-georef: Georeference maps
mm-merge: Merge the layers of several maps into one
mm-info: Inspect map metadata
This makes MOLA maps interoperable with industry-standard tools like CloudCompare, QGIS, and surveying software.
ROS 2 native AND standalone C++¶
Deploy with full ROS 2 integration (Humble, Jazzy, Lyrical, Rolling) or as a standalone C++ application without any ROS dependency: the same modules run in both. This gives you maximum flexibility for edge deployment, Docker containers, or integration into proprietary frameworks.
Multi-sensor, multi-environment¶
LiDAR-only odometry (LO) - no IMU required
LiDAR-inertial odometry (LIO) - with IMU fusion
GNSS fusion - consumer-grade GPS for georeferencing
Kinematics fusion - wheel encoders, vehicle odometry
Validated across urban driving (KITTI), agricultural environments (GreenBot), indoor (warehouses, buildings), outdoor (forests, campuses), and aerial (drones).
Academic rigor, production ready¶
MOLA is backed by peer-reviewed publications in top venues:
IJRR 2025: A flexible framework for accurate LiDAR odometry, map manipulation, and localization
RSS 2019: A Modular Optimization Framework for Localization and Mapping
Benchmarked on KITTI (0.6% translation error), and ready to run on public datasets such as MulRan, Newer College, Oxford Spires, Hilti 2022, GrandTour or VBR (see Running MOLA on a dataset). This is not a paper prototype - it is actively deployed in real-world robots.
Pre-built for amd64 and arm64¶
Binary packages are available from the ROS build farms for both amd64 and arm64 (Jetson, Raspberry Pi). MOLA Pro subscribers additionally get access to a private apt repository and Docker images for streamlined deployment.
Use cases¶
MOLA is used across a wide range of industries and environments: