Exwayz SDK

Component
Delivery
When to use
Mapping
Linux CLI executable
Run offline once to build your reference map
Localization
ROS package · RTMaps module
Run on-robot in real time against the map

Installation

Linux

Releases ship as .deb (recommended) and .tar.gz. The .deb package handles dependency and version management automatically.

sudo apt install -fy <path-to-exwayz-release>.deb

Installs to /opt/exwayz/exwayz_3d_mapping.

License

Activate with the token provided at delivery:

exwayz_license -q <token>
How it works

Run Mapping once offline to produce the reference map, then deploy Localization on your robot. The two components are independent: Mapping can run on a workstation, Localization runs on your embedded hardware in real time.


Pipeline

The mapping pipeline is a sequence of CLI executables run offline. Choose the workflow that matches your environment.

Indoor
01 SLAM Compute odometry trajectory from raw LiDAR data
exwayz_slam --bag <path/to/recording.bag> --bag-lidar /lidar_topic \
  -p narrow -o ./slam_session --no-date
02 Loop closure Detect revisited areas and correct trajectory drift
exwayz_loop_closure \
  -i ./slam_session/frames \
  -t ./slam_session/traj_odometry.ply \
  -p Odometry -d 5.0
03 Merge Accumulate frames into a dense point cloud map
exwayz_merge_cloud \
  -i ./slam_session/frames \
  -t ./slam_session/1_loop_closure/traj_loop_closure.ply \
  -o ./map_indoor.las
Outdoor
01 SLAM Compute odometry trajectory from raw LiDAR data
exwayz_slam --bag <path/to/recording.bag> --bag-lidar /lidar_topic \
  -p wide -o ./slam_session --no-date
02 Loop closure Detect revisited areas and correct trajectory drift
exwayz_loop_closure \
  -i ./slam_session/frames \
  -t ./slam_session/traj_odometry.ply \
  -p Odometry -d 5.0
03 Georef Fuse GNSS and loop closure observations into a georeferenced trajectory
exwayz_georef \
  -t ./slam_session/traj_odometry.ply \
  -g <path/to/gnss.ply> \
  -l ./slam_session/1_loop_closure/lc_observations.yamlc \
  -s 1.0 -r -o ./traj_georef.ply
04 Merge Accumulate frames into a georeferenced dense point cloud
exwayz_merge_cloud \
  -i ./slam_session/frames \
  -t ./traj_georef.ply \
  -o ./map_outdoor.las

ROS / ROS 2

Exwayz ships as a native ROS package. It subscribes to your LiDAR topic and publishes localization outputs directly into your TF tree.

ROS 1 Noetic ROS 2 Humble ROS 2 Jazzy ROS 2 Kilted
exwayz::SLAM LiDAR Odometry
TopicTypeDescription
IN sensor_msgs/PointCloud2 Input point cloud
OUT /exwayz/slam/odometry nav_msgs/Odometry 6DoF relative pose
OUT /exwayz/slam/pointcloud2 sensor_msgs/PointCloud2 Motion-compensated point cloud
OUT /exwayz/slam/status exwayz_msgs/ExwayzStatus Algorithm status
exwayz::map-based localization Real-time Localization
TopicTypeDescription
IN sensor_msgs/PointCloud2 Input point cloud
IN nav_msgs/Odometry Initialization pose
OUT /exwayz/relocalization/odometry nav_msgs/Odometry 6DoF absolute pose
OUT /exwayz/relocalization/odometry_local nav_msgs/Odometry 6DoF relative pose to map
OUT /exwayz/relocalization/navsatfix sensor_msgs/NavSatFix Geodetic coords (lat, lon, alt) if map CRS is set
OUT /exwayz/relocalization/pointcloud2 sensor_msgs/PointCloud2 Motion-compensated point cloud
OUT /exwayz/relocalization/status exwayz_msgs/ExwayzStatus Algorithm status
SVC /exwayz/relocalization/reinit std_srvs/Trigger Trigger reinitialization
SVC /exwayz/relocalization/enable_local_mapping std_srvs/Trigger Start local mapping
SVC /exwayz/relocalization/disable_local_mapping std_srvs/Trigger Stop local mapping and clear map

RTMaps

Exwayz is available as an RTMaps package, enabling seamless integration into Intempora-based ADAS and robotics data pipelines. The package exposes Mapping and Navigation as RTMaps components: drag, connect, run.

RTMaps 4.x
Exwayz RTMaps nodes diagram

Available on the Intempora platform. Visit the AI Store of Intempora →

Supported LiDARs

Exwayz is sensor-agnostic. Any 3D LiDAR producing a standard point cloud is supported. Below are validated manufacturers. Additional models available on request.

Ouster
Hesai
RoboSense
Livox
Innoviz
Seyond
Aeva
Velodyne
Sick
Valeo Scala

Using a different sensor?

Robot platforms

Exwayz runs on any platform that carries a 3D LiDAR. No form factor constraint, no proprietary middleware dependency.

Autonomous trucks & tractors
Port logistics, industrial sites
AMRs & AGVs
Warehouses, manufacturing floors
Autonomous shuttles
Urban mobility, campus transport
Inspection robots
Patrol, surveillance, multi-floor
Mobile mapping systems
Survey vehicles, backpack scanners
High-speed vehicles
Autonomous racing, test vehicles

OS & hardware

Examples of OS & hardware supported.

Ubuntu 20.04
Focal Fossa · LTS
Supported
Ubuntu 22.04
Jammy Jellyfish · LTS
Supported
Ubuntu 24.04
Noble Numbat · LTS
Supported
Jetson Nano / Orin
NVIDIA edge hardware
Supported
Rubik Pi 3
ARM64 edge compute
Supported
Custom hardware
ARM64 / x86_64
Ask us

Frequently asked questions

LiDAR compatibility & sensor setup
Is it compatible with my LiDAR?

Exwayz is sensor-agnostic. Any 3D LiDAR producing a standard point cloud is supported. Validated manufacturers include Ouster, Hesai, RoboSense, Livox, Innoviz, Seyond, Aeva, Velodyne, Sick, and Valeo Scala. If your sensor isn't on the list, additional models are available on request.

What's the recommended mounting or tilt configuration?

There is no strict constraint on LiDAR mounting. The SDK handles any extrinsic calibration as long as the LiDAR provides sufficient 3D point coverage of the environment. Our engineers can advise on optimal tilt angles depending on your use case (indoor, outdoor, mixed).

Does it support multiple sensors?

Yes. Exwayz supports extrinsic calibration between sensors and multi-LiDAR fusion. Multiple sensors can be combined to extend field of view or improve coverage in complex environments.

Integration & technical stack
Do you support ROS? ROS 1 or ROS 2?

Yes. Exwayz ships as a native ROS package compatible with ROS 1 (Noetic) and ROS 2 (Humble, Jazzy, Kilted). It publishes pose, odometry, and status topics out of the box and integrates directly into your TF tree.

Can it run on CPU-only or lightweight hardware?

Yes no GPU is required. The algorithm is designed to run on embedded hardware: NVIDIA Jetson Nano and Orin, Rubik Pi 3, and any x86_64 or ARM64 machine meeting the minimum requirements (4 cores, 4 GB RAM).

How does it integrate with a stack that doesn't use ROS?

For non-ROS environments, Exwayz provides bridge nodes and a standalone executable interface. Data can be handed off via CSV, UDP socket, or direct API calls. Bridge, C++ and Python APIs are coming soon. to discuss your specific integration needs.

What data format do you need?

Exwayz accepts ROS bag (.bag), PCAP, and live UDP socket streams.

Ready to integrate?

Access the full SDK documentation or talk to our engineers directly.