Gas leaks have caused many tragic industrial accidents throughout history. Once detected, a gas leak is often difficult to pinpoint in large facilities or industrial sites. Typically, emergency personnel risk their lives using gas sensing hardware to identify the problem, but a new project from a global team of researchers might take everyone out of harm’s way.
A team of researchers from Harvard University, the University of Barcelona, and Delft University of Technology in The Netherlands have created a swarm of tiny drones that can detect and find gas sources autonomously, even in cluttered indoor environments.
Drones seem ideal for the application, but the challenge was to make artificial intelligence small enough to fit in tight computational constraints. To do so, the team turned to nature and developed Sniffy Bug, a new particle swarm optimization (PSO) navigation algorithm based on the motion of bird flocks. Sniffy Bug was installed on modified CrazyFlie nano quadcopters that measure 12 cm (a little less than five inches) in diameter and weigh a mere 37.5 grams.
If none of the drones sense gas, they spread out, avoiding obstacles and each other. When a drone senses gas, it notifies the swarm, which works together to find the source as soon as possible.
The algorithms can be used for more than detecting gas leaks in buildings. Other applications include the detection of diseases and pests in greenhouses and even finding methane on Mars.
The drones aren’t ready to be deployed just yet as the researchers still need to make improvements to the drones’ navigation as well as their ability to locate gas sources, but one day the Sniffy Bug could be another tool facilities use to keep employees safe and the business up and running.
