Robot manual navigation method based on single-person gestures and voice information

Abstract

The invention discloses a robot manual navigation method based on single-person gestures and voice information. The method includes: collecting human skeleton joint point information and voice information in a current environment through Kinect equipment, and figuring out skeleton information integrity and voice information intelligibility; if the skeleton information integrity is not smaller than a skeleton information integrity threshold, using gesture identification to perform robot manual navigation; if the skeleton information integrity is smaller than the skeleton information integrity threshold and the voice information intelligibility is not smaller than a voice information intelligibility threshold, using voice identification to perform robot manual navigation; if the skeleton information integrity is smaller than the skeleton information integrity threshold and the voice information intelligibility is smaller than the voice information intelligibility threshold, outputting no navigation instructions. The robot manual navigation method has the advantages that gestures and voice are combined to be applied to a robot control system, and accordingly the method is high in practicality, high in accuracy and good in instantaneity.

Description

technical field [0001] The invention relates to the field of robots, in particular to a robot manual navigation method based on single-person gestures and voice information. Background technique [0002] With the development of science and technology, human-computer interaction has gradually shifted from computer-centered to human-centered. Gestures and voice, as people's daily communication methods, are natural, intuitive and clear. Therefore, gesture and voice have become the mainstream of human-computer interaction, and are widely used in remote control, virtual reality, medical diagnosis and other fields. [0003] In recent years, various gesture recognition and speech recognition algorithms have emerged in an endless stream. The more popular gesture recognition algorithms are DTW and HMM. But the recognition rate of DTW algorithm is low, and the robustness of HMM algorithm is poor. Algorithms for speech recognition require training software to improve recognition...

AI Technical Summary

Problems solved by technology

When the gesture recognition algorithm is used alone, the recognition rate of the algorithm will be affected by the lighting, complex background and the integrity of the skeleton information; when the speech recognition algorithm is used alone, it will be affected by the clarity of the speech

[0004] Analysis of the current recognition methods, the performance of the method that only uses gesture recognition is easily affected by lighting, complex backgrounds, and the integrity of skeleton information; the method that only uses speech recognition is easily affected by speech clarity, so it is necessary to provide a new method to solve the above problems

Images