SLAM Reference Software
Soon we will be providing free to download reference solutions so you can try our Spatial AI software for free. If you would like to be the first to know when these are available please sign up to our early access list.Sign up for early access
With nothing but two cameras and a mobile phone grade inertial sensor, SLAMcore:Position is able to track to centimetre accuracy.
Early testing shows it is best in class for accuracy, robustness and computational efficiency. The system can run in both odometry mode and full SLAM mode with loop closure, running on anything from a Raspberry PI and up.
Building on the accurate tracking provided by SLAMcore:Position, SLAMcore:Map uses depth information to create a 3D dense map of the world.
Using only the CPU, our mapping solution allows customers to create surface mesh or global occupancy maps in real-time without the need for post processing.
Our custom neural network can be trained on any labelled dataset, efficiently identifying and segmenting objects of interest.
Combined with our position and map software, it can enable improved tracking and more rich maps, through identifying and positioning dynamic objects.
What you get
We appreciate that accurate, real time positioning and mapping are essential for autonomous robot and drone operation and complex, intelligent machine behaviour.
- ‘6 degrees of freedom’ pose calculated in real time
- ‘Relocalization’ and ‘Loop Closure’ features increase accuracy over large distances
- 3D surface mesh along with local and global occupancy map
We use tightly coupled sensor fusion to provide reliable performance even in challenging operation modes.
- Reliable performance, despite bumps, drops and abrupt or jittery movement
- Multi-sensor approach operates well even in the presence of dynamic objects
Easy to use
We have made sure that our software is easy to use, customize and integrate to your existing software stack.
- Designed to seamlessly integrate with ROS1 and ROS2
- Easy-to-use C++ API
We are a software company, working with off-the-shelf, commercially available and affordable hardware to provide affordable, high quality Spatial AI.
- Does not rely on expensive sensors
- Designed for affordable global shutter cameras and inertial measurement units
- Capable of running on low-cost processors from Raspberry Pi up