The Grid thinks, the machine acts
Czar Grid serves perception, decisions, and adaptation from inside the building. The machine keeps control, motion, stops, and safety. BramHA is being designed for the machine's side and leaves the boundary where it is.
· 4 min read
IN SHORT
Czar Grid is an inference pool that serves perception, decisions, and adaptation to every machine in a facility over the site's network. The machine keeps control, motion, stops, and safety. BramHA is being designed for the machine's side of that line. It does not change the Grid's boundary: the Grid is not a controller and never sits in the loop.
Element 31 already has a product for the heavy AI work in a facility. Czar Grid is several Czars working as an inference pool, serving every machine in a facility, from any vendor, over the site's own network.
BramHA is a separate program, in development, for the other side of that link. No chip exists. This article describes how the two relate and where the line between them runs.
What each side does
A Czar Grid serves perception, decisions, and adaptation from inside the building. A machine sends what it sees and asks what it means or what to do next. The pool answers.
This is the dense, power-hungry part of a robot's work. Placing it in a pool means it can be shared across many machines and changed without opening any of them. It stays inside the building, on the site's own network.
That is thinking, in the loose sense the title uses. It is heavy, and it can take a moment.
The machine acts. It holds its joints, moves them on time, and stops when it must. This work cannot wait for an answer from across a network, and it cannot fail when the network does.
So it stays on the machine. Control, motion, stops, and safety are the machine's own, as its vendor built them.
The boundary, verbatim
Element 31 states the Grid's boundary in fixed words:
Safety, stops and motion stay on the machine as its vendor built them. Czar Grid serves perception, decisions and adaptation over the site's network; it is not a controller and does not sit inside any real-time or safety loop.
Everything about BramHA has to be read with that statement in force. The silicon program does not amend it, soften it, or work around it.
Where the silicon sits
The boundary says what stays on the machine. It does not say what the machine's electronics are made of. That has always been the vendor's choice.
BramHA is being designed as one option for that choice. Its four jobs are all on the machine's side: a real-time controller on which the loop closes, drive for a joint or a motor, reading position, current, and force where they are measured, and a hardware root of trust so that a machine can prove what it is. The chip on the robot's side describes them.
The first design is deliberately small: a real-time controller with a root of trust. It is not a vision processor or an AI accelerator, because that work belongs to the pool.
So the two do not overlap. The Grid serves what is heavy. The silicon is intended for what is strict.
What the silicon does not change
Three things stay as they are.
The Grid still never sits in the loop. Having Element 31 silicon inside a machine would not make the Grid a controller. The link between the machine and the pool carries questions and answers, not control.
The silicon is not a safety system. The chip may run a vendor's control loop, but the safety case is the vendor's. No safety or functional-safety certification is claimed. Where safety lives covers this.
Neither requires the other. A Grid serves machines from any vendor, built with any electronics. The silicon, for its part, is intended as a part a robot maker builds in, and it would do its jobs whether or not a pool is present.
One connection is worth noting. The silicon's identity job is intended to let a machine prove what it is and what it is running before a site lets it on the network. That network is the one the pool is reached over. How a site would use the proof has not been published.
Two programs, one line
The simplest way to hold the relationship is as one line with two sides. On one side is the building's pool, which thinks. On the other is the machine, which acts. Element 31 has a product for the first side and a program in development for the second.
The difference in status matters. Czar Grid is a product. BramHA is not one. No chip has been taped out, the specification of the first design is not published, the process and the foundry have not been announced, and nothing can be ordered.
Does BramHA replace the Czar Grid? answers the most common question about the two in brief.
Questions this raises
- Does BramHA put the Grid inside the control loop?
- No. The Grid still never sits in the loop. The silicon is a part a robot maker may build into the machine, and the loop is intended to close on that part, on the machine.
- Does a Czar Grid need machines built with BramHA?
- No. A Czar Grid serves every machine in a facility, from any vendor, over the site's own network. BramHA is in development and no chip exists.
- Who is responsible for safety when both are present?
- The robot's vendor. Safety, stops, and motion stay on the machine as its vendor built them. The Grid is not a controller, and the silicon is not a safety system.
KEEP READING