ROBOTICS · ADAPTIVE LOCALIZATION
Real-time SLAM and odometry that drops into existing ROS 2 stacks, compiles to native C++, and outpaces GTSAM by up to 3.8× — without changing the surrounding code.
Request a demo
Talk to our team
BEFORE · GTSAM
Slower iterations
Cost climbs with each new factor. The graph grows faster than the solver clears it, and the trajectory falls behind the robot.
AFTER · OURS
3.80x faster
Same problem, same correctness, less compute. The optimizer keeps pace with the sensors and the trajectory stays where the robot is.
Native C++
Compiles end-to-end to C++ with zero-copy transport when configured. No Python overhead in the hot path; nothing between the sensors and the solver.
Drop-in
Integrates natively with existing ROS 2 stacks. No rewrites, no Python shims, no dependency hell — swap the package, rebuild, run.
High accuracy
The math is tight; the trajectory holds when the environment doesn't.
Faster solver
Up to 3.80× peak speedup on sequential build at N=150, 1.71× on loop closure, and ~2× across typical incremental optimization compared with GTSAM.
A new era for robotics
İsmail Şenöz
CTO
Robotics needs to be adaptive by nature. The ideas and research into this field have a long and rich history. And that is exactly what we are improving upon.
“Robotics is one of the most exciting areas to work in. Our goal is to build systems that not only focus on a single task, but that handle the whole loop. From perception to action.”
Build something that moves.
Get in touch to discuss your use case.
Request a demo
Get in touch