We build autonomous machines that work on real floors and real factory aisles — engineered around ROS 2, LiDAR SLAM, sensor fusion, and honest field validation.
Where the work stands
Autonomy is a stack, not a feature. These are the layers our robots are engineered on — from sensing to planning to machine-level intelligence.
The robot software stack runs on ROS 2 so drivers, planners, and system nodes stay modular, testable, and portable across platforms.
LiDAR-based mapping and localization let the robot build a floor map, locate itself inside it, and return to its dock without an operator.
LiDAR, IMU, wheel odometry, and optional vision are fused so the robot keeps its position when surfaces, lighting, or floor layout change.
Vision models help distinguish obstacles, furniture, and people from false readings — so planning decisions are based on what is actually on the floor.
Global and local planners steer the robot through doorways and around people, with recovery behaviours when the path is unexpectedly blocked.
A self-mapping, low-noise floor-cleaning robot designed for industrial and commercial floors.
CleanBot is our first autonomous platform. It learns the layout of a space, plans efficient cleaning routes, and returns to its charging dock — no tether, no dedicated operator. Repetitive floor cleaning consumes thousands of operator hours a year in offices, factories, and institutions across Bangladesh; CleanBot is engineered to take on that cycle.
CleanBot is in active prototype validation on real floors. We are not shipping units yet, and we will not take pre-orders until field trials clear.
Prototype · In active R&D
CleanBot — mapping a real floor
Each stage has an exit criterion. Nothing ships to customers until the previous stage is met.
Now
Next
After
These are planned directions, not commitments. They will open up as the core CleanBot platform matures.
Compact AMRs for moving bins, parts, and payloads inside factories and warehouses — planned once the core autonomy stack matures.
Scheduled inspection rounds for facilities that run continuously — visual checks, environment logging, and report generation without a walking patrol.
Productive cells that combine compact robot arms, vision systems, and safety interlocks for repetitive production tasks.
Development logs, navigation testing, and field validation notes — published as the work happens.
Tell us about the space, the duty cycle, and the outcome you need — we will scope the practical first step.
