A controller at every joint
Motion and sensing happen at the joint. BramHA is being designed to put drive and measurement there too, so that less wiring runs along the machine and one part can be repeated.
· 4 min read
IN SHORT
Two of the four jobs BramHA is being designed for sit at the joint: motion, meaning drive for a joint or motor, and sensing, meaning reading position, current, and force where they are measured. Putting both at the joint shortens the path between measuring and acting, reduces long wiring, and lets one part be repeated. No chip exists.
A robot is a set of joints, and each joint is a small machine of its own. It has a motor to drive and measurements to take. Two of the four jobs BramHA is being designed to do sit there: motion and sensing.
The program is in development and no chip exists. This article describes why those two jobs belong at the joint and what the program intends for them.
The work happens at the joint
A motor turns current into force. Driving it means switching power into the motor in the right pattern, at the right moment, many times over. That is the motion job: drive for a joint or a motor, at the joint.
Driving a motor well depends on knowing three things. Where the joint is. How much current is flowing. How much force the joint is exerting or meeting. That is the sensing job: reading position, current, and force where they are measured.
The two jobs are halves of one cycle. The measurements tell the controller what to correct, and the drive makes the correction. A control loop that closes on the machine describes that cycle.
Drive and measurement in the same place
A measurement is most useful when it is fresh and clean. A signal that is measured at the joint and read at the joint travels a short path. It has less distance over which to pick up electrical noise, and less delay before it is acted on.
The same holds in the other direction. A drive that sits beside its motor switches power over a short path instead of a long one.
So the program's intent is to put drive and measurement where the work happens. This is not a claim about performance. There are no figures for the silicon, and no chip has been built to measure.
Fewer long wires
Consider a machine whose drives all sit in one cabinet. Every joint then needs its own power lines and its own sensor lines running from the cabinet, along the arm, through every joint that comes before it.
Those cables bend each time the machine moves. Cables that flex are a familiar source of wear and faults, and they take up room in parts of a machine that have little to spare.
With drive and measurement at the joint, what runs along the machine can be reduced toward power and a link between the joints. There is less to route, less to flex, and less to trace when something goes wrong.
This is not a new idea. Robot makers have placed electronics in or near joints for a long time. The program's aim is narrower: a part for the job, not a new architecture.
One part, repeated
One joint is much like the next. The motor may be larger or smaller, but the job is the same: drive, measure, report. That makes a joint part a good candidate for one design that is repeated across a machine.
A repeated part has practical advantages. There is one design to verify and one body of firmware to maintain. A robot maker has one part to understand.
It also suits the way the program has chosen to build. A joint's work is switching power and taking measurements, not dense computing. Jobs like these do not need the leading edge of chip manufacturing, and mature process nodes are enough for them. Why robots do not need the leading edge makes that argument.
What is decided and what is not
The first design is a real-time controller with a root of trust. It is deliberately small, and a small first design does little.
Motion and sensing are among the four jobs the silicon is being designed to do. How those jobs are divided across parts, and what a joint part would contain, has not been published. The specification of the first design is not published either.
What will not change is who owns the machine's design. The silicon is intended as a part a robot maker may build in. How a machine's joints are arranged, how they are controlled, and how the machine is made safe remain the vendor's design and the vendor's responsibility. The silicon is not a safety system, and no certification is claimed for it.
Nothing can be ordered. No chip has been taped out, and the process and the foundry have not been announced.
Questions this raises
- Is the first E31 design a joint drive?
- The first design is a real-time controller with a root of trust. Motion and sensing are among the four jobs the silicon is being designed to do, and how those jobs are divided across parts has not been published.
- Is putting electronics at the joint a new idea?
- No. Robot makers have placed drive electronics in or near joints for a long time. The program's aim is a part for that job, not a new architecture.
- Who decides how a machine's joints are built?
- The robot maker. The silicon is intended as a part a vendor may build in. The machine's architecture, its control, and its safety remain the vendor's design and the vendor's responsibility.
KEEP READING