Stial has built a “new generation of intelligent polishing robot AI IoT cloud platform – Mogu Cloud System”, which is based on deep learning, autonomous machine learning, reinforcement learning, domain data augmentation and domain generalization technologies to perfectly solve the problem of workpieces in different fields. Repeated design and programming can shorten the project process debugging cycle from more than one month to one day.
First, let me say that today’s robots can have certain cognitive abilities.
1) The cognitive ability of the machine has changed from artificial definition rules to autonomous definition rules, from stacking machine intelligence in a complex sequence of rules to only assisting the machine in cognition of the conclusion of things, and this conclusion is generated by Processes and rules are self-derived by the machine.
This is the core machine autonomous learning part of the Mogu Cloud created by our company.
2) Completely unsupervised learning. The machine already has a cognitive structure similar to the human brain, but does not have a sufficient knowledge base system, or only has part of the basic cognitive abilities. In subsequent upgrades and expansions, humans do not need to participate in formulating the learning Methods, defined rules, and machines can autonomously build their own cognitive and learning capabilities in the vast amount of big data.
The artificial intelligence we are talking about today is actually the neural network and deep learning in machine learning. This is the ultimate goal of robots, and it is also the capability of robots that have reached the pinnacle of design.
There is another concept that needs to be understood. What is explicit knowledge and what is hidden knowledge
The first thing you think of must be various theorems and formulas, which are recorded in various media such as books, newspapers, and the Internet. Such knowledge is called “clear knowledge”.
What is dark knowledge:
Today, machine learning is a kind of dark knowledge, which is essentially the correlation between large amounts of data. Because the system is too complex and has too many variables, it has far exceeded the cognitive capabilities of human beings. But the more variables there are, the better machine learning becomes. Countless machines will work tirelessly day and night, and we will soon discover that the dark knowledge newly discovered by machines will accumulate rapidly. This is the most revolutionary change that artificial intelligence and machine learning have brought to us.
Assume that our polishing robot, after managing thousands of project experiences, stores a large amount of data related to polishing force control. The larger the amount of data and the more variables, the more accurate the machine will be. This is dark knowledge.
Stial’s mushroom cloud system is built with a large amount of data to achieve intelligent and polished knowledge that personnel do not possess. Perfectly solves the feedback of grinding force control of workpieces in different fields.