What platform is needed for your software: Linux, Windows, Unix, etc.?
Windows NT/2000/XP
If you require, the UMNICK Digital Brain can be easily adapted for Linux/Unix, but at the moment we do not have this version.

What language is the UMNICK Digital Brain written in?
In Visual C++.

The UMNICK Digital Brain - is it DLL, EXE or a simple source code?
A selection of DLL, at the moment we are working on computer based apparatus device (PCI card, microcontroller)

What type of hardware?
Intel Pentium III and higher, or compatible hardware (AMD, Transmeta, etc).

On what scientific research is your product based?
Our product is based on the work of a number of neuro-physiologists and cyberneticists from the last century.
At the moment research using a similar theoretical foundation is ongoing at the MIT laboratory (http://www.ai.mit.edu/projects/humanoid-robotics-group/), the Artificial General Intelligence Research Institute (AGIRI - www.agiri.org) and at a number of independent private companies. However, according to our information, at the moment we are the only company that has managed to complete this development and to actually produce a commercial product.

Does your software manage things such as relay, motor, hydraulics?
The UMNICK Digital Brain is designed to manage arbitrary objects, including relay, motor, and hydraulics.

Can several processors work together (one responsible for vision, while another deals with decision making)?
They can. The architecture of the Digital Brain is designed to work with several processors or computers in a local network.

How is the UMNICK Digital Brain distributed to various computers, perhaps via multiple video channels (let's say 10)?
An abstract model: "many input sensors, many output actuators," makes it possible to implement any possible configuration of the system, including to easily distribute Digital Brains to individual computers in a network. Two options are possible:
1. Strong interaction (central nervous system of the mammal) - adapters between Digital Brains are strictly programmed and it is impossible to change the configuration without reprogramming the System. This is convenient for use when setting up individual robots.
2. Weak interaction (anthill) - each individual computer has a Digital Brain and it interacts with other Digital Brains only through its sensors. Reconfiguration may be constant, without reprogramming - convenient for use when creating a team of robots.

In both cases, all will work as ONE distributed UMNICK Digital Brain.

How many "inputs/sensors" can your software support?
2**63, in other words - as many as memory and computing resources allow, to ensure work in real time.
A Digital Brain server may be set up:
- On one processor.
- On several processors (MPI).
- On several separate computers in a local network.

Does your program replace many hours of coding?
The UMNICK Digital Brain is an attempt to get by without intermediaries between machines and people, i.e. to reduce the volume of coding or to do away with coding altogether. But it is not yet possible to connect a Digital Brain to a machine without coding - modern devices are not capable of this, it is necessary to at least write drivers for the machines.

How does Umnick behave in situations that it has not studied before? How does Umnick know what is "relevant" in a scene?

The Umnick analyzes the entire image; it does not have the concept of importance at the stage of analysis and decision-making - any small detail that changes its decision may be "important."
The concept of importance becomes an issue at the stage of "forgetting" - the UMNICK wipes its memory in such a way as to retain only those fragments of information that were really successful in reaching decisions.
Consequently, the concept of importance is actually used in reaching decisions, albeit indirectly.

As the Digital Brain does not have an understanding of ‘importance,’ or ‘priority,’ how is it able to react quickly and correctly, in dangerous situations for example?

Several Digital Brains can be configured into one Digital Brain. Each of these can be responsible for a task. For example, one will make strategic decisions while the other constantly monitors safety systems in dangerous situations.

Consequently, the configuration of the end System may consist not of one Digital Brain, but a whole series of Digital Brains working as one entity?

Absolutely right, this is yet another competitive advantage of the UMNICK Digital Brain. It’s like a LEGO; everything depends on the list of tasks that the System needs to resolve.

What is the right way to choose the necessary configuration of a System?

Unfortunately, at the moment only UMNICK specialists may provide these services. However, we are preparing materials for special courses where we will teach the most effective use of the UMNICK Digital Brain, both to our Distributors and Clients, and to anybody wishing to learn about the UMNICK Digital Brain.

For example, during the training period there was a cloud in the sky, and this was interpreted by UMNICK as an important detail. If the cloud is not there when work begins in reality, is it possible that the UMNICK might not be able to identify the scenario and might do something unexpected?
If the UMNICK was trained at a moment when there was the constant presence of a cloud in the sky, then in fact it will form only ONE option of behavior: "with a cloud in the sky." If the cloud is absent, then the UMNICK will act in accordance with the ONLY scenario known to him - a mechanism for abstracting unimportant details will come into effect. If the UMNICK was trained with clouds and without clouds, or with different clouds in the sky, then it will form a multitude of different behavioral options, possibly millions (depending on memory size). In reaching a final decision, only ONE BEST option will be chosen. Taking into account signals from other sensors, and not just from video cameras, it is impossible to say in advance which option will be chosen.
The main thing is that, in any case, the UMNICK is NOT able to do anything unexpected - all of its decisions are connected by a context, it can only behave in accordance with the context, whatever the signals from the sensors. In other words, if the UMNICK is trained then it by definition knows how to behave in this situation, as any situation will in some way be similar to a situation already known to the Digital Brain. For example, one pixel of an image is ALWAYS similar to all other images. All chess games are ALWAYS similar to each other - the difference is in the details. All Chinese are similar - it is necessary to remember many Chinese to start to be able to differentiate them from one another. A similar rule applies to all other nations. For the Chinese, all Russians are alike. For black people - all whites, for whites - all black people, etc.

Do you have an existing 'standard line' of products that your company sells? If so what are these products, or do you provide only project consultation and software for concrete projects?
The company Umnick supplies only one product - the UMNICK Digital Brain.
This product is sufficiently general and, in our opinion, it is sufficiently easy to integrate it into any existing product/solution, or use it to create a new line of products.
On the other hand, the UMNICK Digital Brain requires our consultation for use, because it has no analogues on the world market.

Do you have any special requirements (certain protocols, video frequency) for the video camera interfaces?
Standard video cameras, up to 30 frames per second.

How is your product applied to managing autonomous systems?
The purpose of the Digital Brain is to control arbitrary objects. For this it has several mechanisms:
1. The ability to recognize arbitrary N-dimensional patterns, taking contexts into consideration.
2. An abstraction mechanism - the Digital Brain gets rid of all the details of the pattern that are not of importance for reaching a decision.
3. A mechanism of acting in accordance with similar situations - the Digital Brain applies this for accelerated training, i.e. a lot fewer training samples are required to resolve a set task.
4. A mechanism for forecasting the context (future) - the Digital Brain uses this to plan its actions N-steps in advance.
5. A decision-making mechanism - the Digital Brain uses this to select the best decision from a multitude of possibilities.
6. Associative memory (targeting by content/information) - the Digital Brain uses this to quickly search for templates of patterns and possible solutions.
7. A mechanism for switching on/off/inactivating new sensors and actuators (similar to Plug and Play) - the Digital Brain uses this to abstract details received from each concrete sensor/actuator, which makes it possible to easily reconfigure entire system without reprogramming or retraining.

For example, the function of recognizing patterns (see point 1 above), on condition a video camera is connected to the Digital Brain, ensures True Color machine vision with the following functions.