What Is Cognitive Robotics?

Cognitive robotics is a discipline that gives robots intelligent behavior by providing them with a processing architecture. This architecture allows robots to learn and infer corresponding actions to deal with a complex world. We can think of cognitive robotics as an engineering branch of embodied cognitive science and embodied embedded cognition.

Cognitive robotics is a discipline that gives robots intelligent behavior by providing them with a processing architecture. This architecture allows robots to learn and infer corresponding actions to deal with a complex world. We can think of cognitive robotics as an engineering branch of embodied cognitive science and embodied embedded cognition.
Chinese name
Cognitive robot
Foreign name
Cognitive robot

1 Core Questions of Cognitive Robot 1

Although traditional cognitive modeling methods treat symbol coding as a way to depict the material world, the idea of translating the material world into these symbolic symbols is still questionable even if it can hold up. Therefore, the perception, behavior and concept of symbolic symbols are the core problems to be solved in cognitive robotics.

2 Cognitive Robot 2 Starting Point

Cognitive robotics considers the cognitive behavior of animals as a starting point in the development of robotic processing of information, which is very different from many previous artificial intelligence technologies. Anticipated robotic cognition capabilities include perceptual processing, attention distribution, anticipation, planning, complex movement coordination, reasoning about other robotic individuals, or their own mental states. Machine cognition enables the behavior of intelligent individuals to be reflected in the physical world (or in the virtual world, in a state that simulates a cognitive robot). In the end, the robot must exist in real life.

3 Cognitive robots 3 learning skills

3.1 Motor Babble Cognitive Robot 3.1 Motor Babble

The robot's preliminary learning technology is called "Electron of Motors". This involves a number of related random complex actions. These actions are generated by the robot through visual and / or auditory feedback. In this way, the robot starts to form a sensory feedback mode and a corresponding motor output mode. These expected sensory feedbacks can then inform the motor control signals. This is considered similar to how babies learn how to find objects or learn pronunciation. For a simple robot system, such as reverse motion may be used to convert the desired feedback (the desired motion result) to the output of the motor, this step can be skipped.

3.2 Imitation Cognitive Robot 3.2 Imitation

Once the robot can control its motor to produce the desired result, then the method of "simulation learning" can be used. A robot monitors the behavior of another agent, and then the robot attempts to imitate that agent. Converting simulation information into desired motor results in complex scenarios is often a challenge for robots. It should be noted that simulation is a high-level form of cognitive behavior, and imitation behavior does not necessarily need to be based on a basic model that reflects animal cognition.

3.3Knowledge acquisition Cognitive Robot 3.3 Knowledge Acquisition

A more complicated learning method is "autonomous acquisition of knowledge": let the robot explore its environment on its own. A series of representative hypothetical goals and ideas are formed.
A more direct exploration mode can be implemented with "curious" algorithms, such as "smart adaptive curiosity". These algorithms usually involve breaking a limited number of categories of sensory input and assigning prediction systems to each. This prediction system can track prediction errors over time. Reducing prediction errors can be viewed as learning. The robot then preferentially explores the species that it has the fastest (or reduces prediction errors).

4 Cognitive Robot 4 Other Architectures

Some researchers in cognitive robotics have tried to use architectures such as (ACT-R and Soar (cognitive architecture)) as the basis for cognitive robotics projects. When building simple and symbolic laboratory data models, these highly modular "symbol processing" architectures are used to simulate operator and human behavior. The idea is to extend these architectures to handle real-world sensory input, which is continuously unfolded over time. What we need is a magical blow to transform the world into symbols.

IN OTHER LANGUAGES

Was this article helpful? Thanks for the feedback Thanks for the feedback

How can we help? How can we help?