DENSO robot architecture

Robot architecture refers to the structural method of information processing and manipulation logic of one or several robots in order to complete the specified target.

1. Differentiation according to function

According to the architecture of functional differentiation, it is attributed to the traditional prudent intelligence in artificial intelligence, embodied in serial dispersion in structure, and asynchronous fulfillment in fulfillment method, that is, information processing and manipulation are realized in the form of "perception, planning, and action". It is typical of the NASR human MtI proposed by the United States China National Space Administration and the United States National Bureau of Standards]. The advantage of this architecture is that the function of the system is clear. The hierarchy is clear and the implementation is simple. However, Shenxing's approach greatly prolongs the response time of the system to external events, and the change in the environment makes it necessary to plan from scratch, thus reducing the performance power. Therefore, it is only suitable for more complex work in a known structured environment.

DENSO robot architecture

2. Differentiation according to behavior

According to the architecture of behavior differentiation, it is attributed to modern reactive intelligence in artificial intelligence, which is embodied in parallel (accommodating) distribution in structure, and synchronous fulfillment in terms of fulfillment method, that is, information processing and manipulation are carried out in parallel in the form of "perception and action". Typical examples are the Subsumption Architecture proposed by R.A. Brooks of the Massachusetts Institute of Technology and the architecture based on MotorSc hema proposed by Arkin. Its first advantage is short performance time, high power and strong maneuverability. However, due to the lack of overall management, it is difficult to adapt to various situations. Therefore, it is only suitable for performing relatively simple tasks in the Mutao environment.

3. Differentiation according to the dispersion of intelligence

According to the architecture of intelligence dispersion, it is attributed to the latest distributed intelligence in artificial intelligence, which is reflected in the structure of scattered dispersion and collaborative fulfillment in terms of fulfillment, which can not only complete the solution of their own local problems independently, but also solve single or multiple overall problems through collaboration. It is typical of the architecture according to the multi-agent. The advantage of this architecture is that it has both the characteristics of "intelligent dispersion" and a consistent coordination mechanism. However, how to reasonably distinguish and coordinate between various agents still requires a lot of research and practice. This architecture has a wide range of applications in many large-scale intelligent information processing systems.

In addition to the above three primary architectures, there are also some improved hybrid architectures, such as behavioral differentiation with feedback links, hierarchical architectures based on distributed intelligence, multi-agent architectures based on function, and so on. However, on the whole, they either lack the flexibility and extensibility of utility modules, or fail to coordinate the careful intelligence and reactive intelligence, or the communication mechanism between the various levels is not perfect.

DENSO robot architecture

Sections viewed

Contact

working hours