Prodrive Technologies logo white Prodrive Technologies logo

Prodrive Technologies and Robovision partner for AI-enabled Automated Defect Classification

Prodrive Technologies and Robovision partner for AI-enabled Automated Defect Classification

SCIENCE PARK EINDHOVEN (July 06, 2023) - Prodrive Technologies and Robovision are partnering to deliver AI-enabled Automated Defect Classification systems. Robovision’s best-in-class AI-ADC system enables operators and domain experts with no AI background to build their own deep learning-based systems quickly and with no code. Prodrive Technologies uses this to supply customers with an optimized turnkey semiconductor defect review and classification system running on its own proven hardware. It is being shown at Semicon West (July 11-13, San Francisco, booth #5645).  

Advanced ICs have hundreds of layers with features down to 3nm, each of which requires a new model based on a unique dataset. Manually developing and training inspection systems can take too long. ADC systems have recently advanced with Deep Learning allowing operators to quickly and accurately train their own AI models to detect and classify defects in real-time. Retraining the models keeps accuracy and quality high as process conditions and faults change. Manufacturers can respond far more quickly to production faults to eliminate them. That improves production yields, raises throughputs, and cuts lead times.  


Machine learning helps address the worldwide shortage of personnel 

AI automates manual work, for example where a microscope image examines the quality of a wafer layer. That helps address the industry-wide problem of finding enough qualified and experienced personnel to operate and maintain complex inspection systems. It also makes error classification more accurate and repeatable by reducing human error.  

Prodrive Technologies in collaboration with Robovision optimizes the system to specific applications, whether a business is exploring AI for the first time or is looking to improve and update existing systems. The NVIDIA-certified GPUs and networking have been tested specifically for contemporary AI workloads. Importantly, the systems are an add-on to existing equipment in the field, which enhances the productivity of these machines and helps to extend their use.