A parking stack you did not have to staff
Add autonomous parking to a programme without standing up a multi-year internal autonomy team, and without betting the launch date on it.
Solutions
Land. Expand. Multiply.
Prove the platform in one demanding category, port it to the next, then let it scale across many. We started with automotive because it has the least tolerance for failure — and the discipline carries into everything after it.
Land · Automotive
Autonomous parking
Parking is a deceptively hard navigation problem. The vehicle has to localize precisely in a multi-storey garage with no satellite signal, transition to open lots and back, read a space it has never seen before, and manoeuvre a large body through tight geometry with pedestrians present — all on production automotive compute, validated to automotive standards.
Solving it end to end is what produced the core the rest of the platform is built on.
Indoor + outdoorCoverage
Continuous localization across garage levels, ramps and open lots.
Automotive gradeValidation
Built and released to the reliability discipline OEM programmes require.
Production computeConstraint
Runs within the cost and power budgets of a shipping vehicle.
Software behind your hardware
Package the navigation layer with your sensing or compute offering and take a complete, validated capability to your OEM customers.
Already in production
Running in production programmes with a leading Chinese automotive OEM, and with a second OEM reached through a global Tier-1 supplier.
Expand · Retail
Smart carts and store robots
A supermarket is the opposite of a parking garage in almost every way: smaller robot, cheaper sensors, tighter aisles, a layout that changes weekly, and people everywhere. Porting the same core into it — rather than writing a second stack — is the test of whether a navigation platform is really a platform.
It is in development now with a leader in retail robotics, on a signed programme.
Aisle-levelPositioning
Accurate enough to guide a shopper to a product, not just a department.
Changing layoutsMaps
Site maps stay current from routine operation as displays and stock move.
Human trafficBehaviour
Motion that shoppers and staff find predictable and comfortable.
The proving ground
Dynamic human environments
Robots that share space with moving people face the hardest version of the navigation problem: the world reacts to the robot, and the robot has to be legible to the world. Master that, and quieter environments become straightforward.
Unpredictable motion
People change direction without warning, cluster, stop dead and appear from behind occlusions. Planning has to assume all of it.
Real-time decisions
There is no time to re-plan from scratch. Decisions have to be made inside a tight latency budget on production compute.
Legible behaviour
A robot that is technically safe but behaves strangely will not be tolerated. Motion has to read as intentional to the people around it.
Multiply · Emerging categories
Where the platform goes next
These are categories the platform is built to serve and where we are actively looking for partners. If you are building in one of them, an early conversation shapes what we support next.
Sport and fitness
Robots that track, pace and interact with people in motion, outdoors and on court — among the most demanding dynamic environments there are.
Security and inspection
Patrol robots covering large mixed indoor and outdoor sites, day and night, with reliable localization across the boundary.
Warehouse and logistics
Movement through structured facilities that are never quite as structured as the drawing suggests, alongside human workers and vehicles.
Hospital and care
Transport and service robots in corridors and lifts, where predictable, unobtrusive behaviour around people is a requirement, not a nicety.
Partners
Who we work with
The platform only works if it fits the ecosystem around it. We build alongside the companies that supply, integrate and deploy robot hardware.
Robot makers and OEMs
License the navigation layer and put your engineering into what makes your robot different, instead of into the part every robot needs.
Tier-1 and component suppliers
Pair your sensing or compute platform with validated navigation software and offer your customers a complete capability.
Integrators and operators
Deploy robots into new sites faster, with mapping, tuning and fleet telemetry handled by a platform rather than per-site heroics.
Building a robot that has to move?
Tell us the embodiment, the sensors and the environment. We will tell you what is proven today and what would be new work.