Event-based machine vision in Industry 4.0 applications

Prophesee SA (formerly Chronocam) has introduced its first commercial implementation of its event-based vision technology for machines. The Onboard reference system is aimed at developers of vision-enabled industrial automation systems such as robots, inspection equipment, and monitoring and surveillance devices. It features a Prophesee-enabled VGA-resolution camera combined with a Qualcomm Snapdragon processor and can be quickly integrated into a production camera system design.

The Onboard reference system provides a guide for developers to implement Prophesee’s neuromorphic vision technology for use in applications including area monitoring, high-speed counting, vibration measurement, or real-time arc welding monitoring. Based on the company’s proprietary technique that uses a bio-inspired approach based on capturing events that change in the field of view of individual pixels (and avoiding the transmission of redundant data), machine vision systems that utilise Prophesee’s image data acquisition and processing solutions are said to save computational power, bandwidth, memory and energy.

The event-based approach is also claimed to enable sensors to achieve higher dynamic ranges than commonly associated with high-speed vision. According to Prophesee, it allows cost-efficient sensors and systems to record events that would otherwise require conventional cameras to run at 10,000 fps and more.

“We are confident that the inherent efficiency of event-based sensing will make it possible for companies to deploy machine vision more widely and no longer have to deal with large amounts of (redundant) data when using cameras at high frame rates to achieve high-speed vision,” said Luca Verre, CEO and co-founder of Prophesee. “The combination of performance and efficiency will help speed up production lines, improve quality monitoring, and facilitate better and safer human/machine collaboration in production environments.”

The solution is also suited for monitoring and surveillance. The vision sensor’s high dynamic range allows operation in difficult lighting environments while producing much less data than a traditional frame-based approach and reducing the computational burden of scene analysis processing.