Autonomous systems.
Decide on board. Learn on the job.
Duplex runs on board robots, drones and vehicles, on the CPU they already carry, beside the autonomy stack that sees and moves them. Each machine decides what is on its list, hands over what it is not sure of or may not decide, and learns from every correction an operator makes, with or without a connection.
Beside the autonomy stack.
Your autonomy stack sees the world and moves the machine. Duplex works beside it, on the same computer: it decides what to do next from what it has learned, brings in your models for what is new, and leaves every movement to the controls and safety systems already there.
A warehouse robot’s cameras pick up something lying across its aisle.
Its own list, and no further.
Each machine makes the decisions on the list your team gave it, and no others. What is new goes to your models, and what is not its call goes to an operator, so a machine never acts beyond what you set.
Its battery runs low far from base, and heading back is on its list, so it turns for home at once.
It acts alone only when it is sure.
Duplex weighs how well each decision is supported before the machine relies on it. When it is sure, and the decision is on the list and within your rules, the machine acts and the reasons are recorded. When it is not, it hands over.
Spraying single weeds is on the robot’s list.
Every correction, learned on board.
An autonomy stack learns when it is retrained and shipped again. Duplex learns from each correction an operator or supervisor makes, on the machine, as the work happens, so the next decision already benefits. Any lesson can be taken back.
After rain, a haul truck heads for ramp 2, which turns slippery when wet.
Out of range, still within its list.
Under water, underground or beyond the last mast, a machine keeps learning and deciding on board, within its list. What is not its call waits for people, and when the link is back, every step is there to review, in order.
08:00: an underwater vehicle leaves its support ship to inspect a pipeline.
What one machine learns, the fleet can use.
What a supervisor teaches one machine can reach every machine you allow, while a site’s own exceptions stay with it. A lesson that proves wrong can be taken back, wherever it was shared.
At site A, a supervisor confirms that robot 7 should wait at a half-open door, not turn back.
The machines it runs on.
Wherever a machine carries a CPU and has work to learn from, Duplex can run beside the autonomy already on board.
Warehouse and factory robots
Mobile robots and arms that learn each site’s exceptions, shift after shift.
Drones
Inspection, survey and delivery drones that decide in the air and hand back what is not theirs to decide.
Farm machines
Tractors, sprayers and field robots that keep working where there is no signal.
Mining and construction
Haul trucks, drills and site vehicles that learn from every supervisor’s correction.
Ports and yards
Yard tractors and terminal vehicles that learn each site’s routes and exceptions.
Ships and underwater vehicles
Vessels and underwater vehicles that work for hours or weeks with no link at all.
Common shapes.
Every deployment is planned with you. Most start from one of these.
One machine, fully on board
Duplex runs on the machine’s own computer, with no outside connection at all.
A fleet with a centre
Each machine decides on board, and a central Duplex in your cloud sees the whole fleet.
Supervised from a control room
Machines decide what is on their list, and operators in a control room take the rest.
A short list first
Start with a few decisions on each machine’s list, and add more as each one proves itself.
Plan your deployment with us.
Duplex runs in a cloud account you own, or on your own devices and machines at the edge, on CPU. We plan each deployment with you. Contact us at hello@vareon.com.