Chapter 465

Although the problem of delay and consistency is the most basic problem in the research and development of intelligent robots, it cannot be ignored.

From instruction delivery to instruction execution, Nebula's intelligent robot can be controlled in about 10 milliseconds.

But that's not enough, it's not enough!

At present, the industry's top level, the latency cannot exceed 5 milliseconds.

Perhaps ordinary users can't notice the difference between 10 millisecond latency and 5 millisecond latency, but this is not the reason why Nebulas Technology does not pursue improvement.

After all, if the company can really control the latency to less than 5 milliseconds, it can be used as another propaganda point.

Let's talk about consistency.

The so-called consistent control is to design a suitable controller to coordinate and control the individual agent, which can continuously adjust and update its own state with the help of the state information of the neighboring agent, and finally make the state of all agents in the system converge.

To put it simply, it means that each component of the intelligent robot completes a unified specified action at the same time.

There is a necessary connection between the latency problem and the consistency control problem.

Officially, due to the existence of delay, the various systems of the intelligent robot cannot reach an agreement.

However, this does not mean that the consistency problem is solved by solving the latency problem.

To achieve consistent control, it is not enough to reduce latency.

"Mr. Lanny, can I take a look at the process parameters of your company's intelligent robots?" Cheng Nuo turned to Mr. Lanny.

Mr. Lenny hesitated for a moment, but then nodded affirmatively, "Okay, I'll have someone call up the data right away." ”

Ten minutes later, a researcher in a white lab coat walked in with a stack of forms.

"President Lanny, here's the data you asked me to print. ”

"Well, just give it to this gentleman. ”

Cheng Nuo took the form and glanced at it with ten lines.

Nebulas currently produces a total of 15 kinds of intelligent robots, and due to the different forms and functions, there is a large gap in the process parameters of each intelligent robot.

But the components are basically similar.

The mechanical body, controller, sensor, and driver are divided into four parts.

For example, the simplest intelligent sweeping robot is composed of a disc-shaped mechanical appearance, a collision sensor, a continuous ranging infrared sensor, an Attmel microcontroller, and a harmonic gear drive.

Cheng Nuo briefly flipped through the table and found that among the 15 kinds of intelligent robots, the only thing that the four parts of the components are the same is the controller part.

All use Atmel microcontrollers, with 4k bytes of ISP flash memory, three-level program dongle, 128X8 bytes of internal RAM, 32 programmable I/0 ports, and 2 16-bit timing/counters.

This kind of single-chip microcomputer could be said to be good a few years ago, but today it can only be said to be average.

Since it is necessary to solve the problem from the root, Cheng Nuo believes that this controller is the key.

"Does your company have the idea of replacing the controller inside the intelligent robot with a single-chip microcomputer? For example...... Motorola microcontrollers?"

In Cheng Nuo's concept, if you can solve problems with money, try to solve them with money.

Mr. Lenny smiled bitterly, "Professor Cheng, I'm afraid you don't know the price of the Motorola microcontroller, just this one-chip microcontroller, I'm afraid the total cost of the other three components is higher." ”

"In that case, we don't have much profit to speak of. ”

"Okay. Cheng Nuo shrugged his shoulders and said indifferently, "Since that's the case, then I can only come by myself." ”

Cheng ...... Professor Cheng, what do you mean?"

"Since the Atmel microcontroller can't meet your requirements, and the reality doesn't allow you to buy a higher quality controller, the only way is to build your own. Cheng Nuo replied very naturally.

But Mr. Lenny heard it as thunder.

He swallowed hard, "Build a single-chip microcomputer, us?"

Cheng Nuo smiled calmly and pointed at himself, "No, it's me, myself." ”

Cheng Nuo raised his hand and kept smiling, "Mr. Lanny, don't be surprised, of course I don't mean to create a new single-chip microcomputer, that is basically impossible, even if you add your company's scientific researchers, it is difficult to do it." ”

"And what do you mean?" said Mr. Lenny.

"Retrofitting on the basis of the Atmel microcontroller. Cheng Nuo replied, "Under the condition that its performance parameters remain unchanged, the purpose of consistent control of the intelligent system is achieved by changing the conduction mode of the internal structure." ”

"This should be difficult, right?" Mr. Lenny is not a scientist and does not know much about it.

Cheng Nuo nodded approvingly, "Well, it's indeed a bit difficult. ”

"How long will that take?" asked Mr. Lenny eagerly.

Time is money, and the big conglomerates are sharpening their knives at the moment, and they can't tell when they will put their butcher knives on the necks of their fattened pigs.

In Mr. Lanny's concept, it will take at least one to two weeks to improve a microcontroller!

Unexpectedly, Cheng Nuo said casually, "Half a day." ”

Burst-!

Mr. Lenny almost vomited out a mouthful of old blood.

Quite a while!

Mr. Lanny's eyes widened with a look on his face of whether you were teased or not.

Then, in a hoarse voice, he said, "Really?"

Cheng Nuo nodded innocently, "It is conservatively estimated that it will take half a day, in fact, if everything goes well, two hours is enough." ”

Mr. Lanny: "......"

If he hadn't known the reality of Cheng Nuo in advance, he would have doubted whether he suspected that he had paid a liar to come back.

But out of awe of Cheng Nuo, he still chose to believe Cheng Nuo, "Then what you need, I'll prepare it immediately." ”

"A computer, a quiet room, scratch paper and pens, by the way, bring another Atmel microcontroller. ”

Cheng Nuo is also welcome.

Soon, everything was arranged.

Cheng Nuo glanced at the time, 9:34, if it was fast, he could still catch up for lunch.

For the improvement of Atmel microcontroller, Cheng Nuo has long had a belly plan.

In fact, simplifying the problem of consistency and expressing it in mathematical language is a fairly common network topology problem.

Using the generalized Nyquist criterion and Gerachgorin's disk theorem in frequency domain control theory, a dynamic model can be established by describing the directed graph.

With Cheng Nuo's current level, although solving this kind of problem cannot be described as hand-to-hand, there is still no big problem in running smoothly.

【...... If the system λi(G(e^jω) is in the unit circular region λi(G(e^jω)∈{θ:θ∈C,|θ-(c+jd)|1/2 is true. 】

[For discrete multi-agent system 3, if for ∨i∈I, there is τ<2(1-Td)/(1-α)π^2Td, then the discrete multi-agent system is asymptotically stable and reaches consistency, then ......]