Virtual Assembly Technology Research Based On Chinese Voice Interaction

At present, the virtual assembly systems usually rely on the force-feedback or somatosensory devices as interactive tools, using some intuitive gestures, such as grab and drag, to complete the assembly of product. However, when some subtle parts need to be assembled, the results are not very desirable.Aiming at this problem, this paper puts forward with the Kinect somatosensory equipment for input, using Chinese voice command recognition to guide the assembly process. The bicycle virtual assembly system based on Chinese voice interaction verified the feasibility of the method. The specific contents of the thesis include:(1) Because of virtual assembly based on Chinese voice interaction involved in virtual environment modeling, rendering, Chinese speech recognition technology, etc. Therefore, the design focus to consider a variety of software communication methods. As a case study of it, a bicycle virtual assembly prototype system is implemented.(2) Design and simulation of virtual assembly environment. Using the3DS MAX to create the scenes, models; Photoshop CS6complete interactive menu design and material texture image processing. Then import the resources to3Dvia Studio Pro to complete the entire scene rendering, assembly sequence planning, etc.(3) In order to reduce the operation error, an assembly sequence planning based on the recommended is proposed. By analyzing the physical connection relationship, hierarchical relationships, and assembly constraint relations at the same level, defines the bicycle assembly priority model. Only when the higher priority parts assembled, the next priority parts of system will be recommended for assembly.(4) Considering the multiple speakers, this paper proposes a Chinese voice command recognition method based on the Kinect sensor, and tested the recognition accuracy of the method through the experiments. By using Kinect sensor to capture user’s skeletal tracking data in front of a certain range, and return the nearest skeleton information; Then, using the coordinate information of the user’s head and linear microphone array to locate the operator, which can identify the user’s voice commands and complete the operation through sending keyboard events.