How can AI target 'zombie' batteries – and make recycling safer? 

Exploding or zombie batteries can cause injury and disruption during the recycling of discarded electronics. Experts at The University of Manchester have used artificial intelligence (AI) technologies to develop a vision-based device to scan electrical junk and detect problem batteries – even in the most damaged equipment. 

The Environmental Services Association warns that too many batteries are carelessly discarded  in bins where they're easily damaged  and can start to burn, creating so-called 'zombie' batteries.

Profesor Hujun Yin of the Department of Electrical and Electronic Engineering has created a tool that helps reduce fire and safety hazards in recycling across items such as laptops, phones and tablets, DIY power tools, and other home appliances.

Putting out fires 

This issue poses a major fire hazard and small fires at recycling centres are a daily occurrence in the UK (on average). As well as being disruptive to the recycling sector, these batteries represent a serious health and safety risk to workers in recycling plants.

Using deep learning technology and images from a moving recycling conveyor belt, Professor Yin’s intelligent vision system is able to recognise zombie batteries, even if deformed or degraded. An in-house prototype has been developed through a Knowledge Transfer Partnership with Benson Components Ltd and successfully tested, using a test rig installed by waste services experts Biffa.

The next step for Professor Yin's AI project involves scale-up and a ready-for-market product, involving an exciting Knowledge Transfer Prtnership between The University of Manchester and Benson Components Ltd, supported by Innovate UK.

Defeating zombies with AI

Businesses across all sectors are becoming increasingly reliant on advanced machine learning technology and automation for making sense of their data, and optimising decision-making, efficiency and performance.

AI (including machine learning and deep learning), data analytics and optimisation techniques have advanced significantly in the last decades, and are now having a revolutionary impact on most industries and businesses – including those looking to mitigate the impact of climate change.

• 

  Professor Hujun Yin

  Professor Yin is a Professor of Artificial Intelligence in the Department of Electrical and Electronic Engineering.

  View Professor Yin's research profile