ANTICIPATE is gold member of the Intel® partner alliance program. With the Intel distribution of the OpenVINO Toolkit, customers can be served more flexibly and faster, especially in the pilot phase.

‍

Introduction to Manufacturing Challenges

Currently, 72% of factory tasks are performed manually, leading to 40% of manufacturing errors being human-caused. Traditional quality controls, typically positioned at the end of production lines, often fail to catch these mistakes early, resulting in increased costs due to late-stage error detection.

‍

AI-Based Assistance System

To address these inefficiencies, we are developing an AI-based system that assists in the automated verification of manual assembly processes. This system utilizes cameras to ensure that assembly steps are completed in the correct sequence with the appropriate components, detecting errors in real time. Immediate corrections at the assembly station can significantly reduce costs related to root cause analysis and subsequent error correction.

‍

‍

Training and Process Optimization

Additionally, this AI system supports training new employees more efficiently, speeding up their readiness for operations. The data obtained on the frequency of various errors in individual process steps can then be used by production planning for process optimization.

‍

Hardware Selection and Cost Management

For hardware deployment, the use of portable edge devices during the pilot phase helps keep costs low, minimizing setup time and investment risks. In contrast, the scaling phase may benefit from centralized computing resources to reduce per-solution hardware expenses.

‍

Overcoming Data Challenges with Intel’s OpenVINO

One major challenge in deploying deep learning in manufacturing is the scarcity of training data, which necessitates significant initial data collection and annotation efforts, elevating pilot costs. Here, Intel’s OpenVINO toolkit comes into play. It simplifies the optimization of deep learning models for deployment on various Intel hardware, saving on hardware and energy costs. OpenVINO's toolkit includes tools for converting and optimizing models, as well as an inference engine for executing these models efficiently on selected devices. Additionally, Intel's Model Zoo offers pre-trained models that help overcome data shortages, reducing the efforts needed for initial model training.

‍

‍

Partnership and Future Prospects

We are excited to collaborate with Intel to enhance our offerings, support manufacturing workers more effectively, and strengthen the role of AI in industrial settings. Our partnership with Intel has already demonstrated a strong focus on achieving results and mutual benefits, highlighted by their responsiveness and technical expertise, as noted by Gennaro Cavallucci, AI ISV Strategic Relationship Manager at Intel Corporation.

‍