AI-based noise suppression service for an immersive human-technology interaction on shop floor with voice assistants

Due to the ongoing digitalization, more and more IT systems are being used for manufacturing planning, control and quality assurance processes. Over time, these systems reach deeper and deeper in the automation pyramid - even down to the shop floor level. Current operating and human-machine interaction concepts, such as keyboards, touch screens, tablets or terminals, require the full attention of the users. The systems must always be operated manually with hands. The hands must be free and clean for these operation. This causes interruptions in work processes, so that enormous potential is lost.

Voice assistant systems such as Siri, Alexa, Samsung S Voice (Bixby) or Google Now represent a new innovative possibility for human-machine and human-AI interaction. They can release enormous potential, especially in the manufacturing area, as they can be used parallel to work processes without work interruptions. Due to the large amount of loud background noise on shop floor, these voice systems cannot be used so far.

Aim of the project is the development and demonstration of an intelligent AI-based background noise suppression service. The service should be provided as an out-of-the-box solution over the AI-on-demand platform as an pre-processing component for voice assistant solutions. The service enables an immersive human-technology interaction via voice commands directly on the shop floor without interrupting work processes.

As part of the experiment, a demonstrator should be set up at our partner MIBRAG. For this purpose, the AI-based noise suppression service should be combined with an existing external natural language processing (NLP) service from the AIREGIO AI solution catalogue. The demonstrator is intended to demonstrate a new front-end technology for interaction with the existing conveyor belt management system as an exemplary representative for many other IT systems.

The USP of the solution is a fully integrated novel AI-based noise suppression approach, which enables voice interactions in loud manufacturing environments. The service is the enabler for new voice assistant based frontend technologies. It could be used in numerous manufacturing applications, such as ERP, PPS, QMS, SCADA, machine control or other automation solutions. The service increases the degree of digitalization, productivity and safety at shop floor level in manufacturing and is an important enabler for industry 5.0.

InfAI has already developed a prototype of the AI-based noise suppression system for MIBRAG and tested it in a laboratory environment. In the experiment, this prototype should be ported as a new stand-alone service to the AI-on-demand platform and combined with an external NLP service from the AIREGIO AI solution catalogue. The service should provide standardized input and output StreamPipes. The development of the noise suppression service should be completed after the end of the fourth project month. In the following four months, the demonstrator at MIBRAG should be set up and evaluated on the basis of the defined use cases. Therefore the service has to be integrated with the existing conveyor belt management system, which was also developed by InfAI. This should be done after the sixth project month. At the end of the project, a real world demonstrator should be available, which enables voice control of the conveyor belt management system on shop floor level without interrupting work processes.