- Can sense up to ten hazardous chemicals including acetone, ammonia and methane
- Opens up a possible union of AI and neuroscience
All this while, we had been dependent on animals with a strong sense of smell such as dogs to sniff out traces of undetectable substances. Although highly effective, this technique comes with its challenges involving fitness and skill maintenance.
Now, researchers from Intel Labs and Cornell University have successfully demonstrated the ability of Intel’s neuromorphic chip, Loihi, to accurately detect smells emitted by hazardous chemicals. This not only will ease the above issue but will also open doors for the convergence of neuroscience and artificial intelligence.
Inspired by human sensory organ
Whenever a living being takes a whiff, the olfactory cells present in the nose get activated, which then send a signal to the brain’s olfactory system where electrical pulses within an interconnected group of neurons generate the smell’s sensation (pleasant or unpleasant).
Similarly, with the help of a neural algorithm that was based on the architecture and dynamics of the brain’s olfactory circuits, researchers trained Loihi to recognise the scents of 10 hazardous chemicals. For this, a dataset consisting of the activity of 72 chemical sensors in response to certain smells was fed into the Loihi’s circuit. The chip immediately learned the neural representation of each sample and recognised it, even in external conditions that prevented it from functioning as desired. In the end, the chip was effectively able to detect chemicals such as acetone, ammonia and methane.
Nabil Imam, Senior Research Scientist in Intel’s Neuromorphic Computing Lab said, “We are developing neural algorithms on Loihi that mimic what happens in your brain when you smell something. This work is a prime example of contemporary research at the crossroads of neuroscience and artificial intelligence and demonstrates Loihi’s potential to provide important sensing capabilities that could benefit various industries.”
Smoke and carbon monoxide detectors are equipped with sensors to notice odours but are not able to distinguish between them. Through the neuromorphic chip, these devices can overcome this problem.
Robots having Loihi can be deployed for environmental monitoring, hazardous materials detection, or for quality control in factories. They could also be used for medical diagnoses where certain diseases emit specific odours.
“My next step,” Imam says, “is to generalize this approach to a wider range of problems — from sensory scene analysis (understanding the relationships between objects you observe) to abstract problems like planning and decision-making. Understanding how the brain’s neural circuits solve these complex computational problems will provide important clues for designing efficient and robust machine intelligence.”
Further research is going on to overcome any complex olfactory signal recognition issues before this becomes a product that can solve real-world problems.