Robot motion skill learning method and system

Abstract

The invention discloses a robot motion skill learning method and system. The method comprises the steps: acquiring a data sample set of human dragging teaching; performing dimension reduction processing on the data sample set based on a principal component analysis method; establishing a variable constraint condition in an implicit space, and screening the data sample set after dimension reduction processing in combination with the variable constraint condition to generate an implicit space data set; performing modeling learning on a hidden space data set by adopting a Gaussian mixture model and a Gaussian mixture regression method, and outputting a robot motion control training model; and predicting the robot motion control training model based on a recurrent neural network, solving a model optimization solution, and converting a model optimization solution into a robot actual control quantity. In the embodiment of the invention, the autonomous learning of the robot motion skills can be realized by using a small amount of human teaching data and considering the inherent constraints of the robot body, so that the generalization ability and programming efficiency of the algorithm are effectively improved.

Description

technical field [0001] The invention relates to the field of robots and artificial intelligence, in particular to a method and system for learning robot motor skills. Background technique [0002] Robot motion skills can realize robot motion planning and motion command generation for a given task, which is the basis of robot intelligence. Aiming at the research hotspot of autonomous movement of robots in complex environments and tasks, how to endow robots with human operation skills becomes the key. Traditionally, off-line programming or teaching programming is usually used to solve the motion task by geometrically describing the motion task and combining the robot kinematics model and interpolation method. However, this method has poor adaptability to complex tasks and difficult task description. , The need for repeated programming for the same type of tasks. [0003] With the rise of artificial intelligence technology, relevant technical personnel propose to extract huma...

AI Technical Summary

Problems solved by technology

This pure data-driven learning method can effectively improve the robot's task adaptability and programming efficiency, but in the process of operation, in order to learn from human's ability to adapt to complex tasks and environments, there are the following shortcomings: (1) Human teaching data is limited, Especially when the position of the robot is not good (such as near the joint limit, near the singularity, etc.), there is no effective training data; (2) a large amount of experimental data is required for collection and labeling, which makes the built neural network structure huge, and the hardware implementation of the algorithm more difficult

Images