“The area we are focusing on is river rescues, basically suicide attempts in inner-city areas, using drone and artificial intelligence to help the search-and-rescue teams to cover the initial window of the rescue and provide them with intelligent information,” says chief executive Eoghan Mulcahy.
The idea first struck Mulcahy as he witnessed a rescue attempt at first hand in February last year.
“It was a specific event that caused the idea to happen. I witnessed someone enter the water in Limerick city and I witnessed the aftermath of the rescue attempt. It was very much a kind of frenzied rescue attempt, because the response was very quick but it was a ground-based rescue team. They were trying their best to pinpoint the location of the person who had entered into the water but is was nearly an impossible task because the river flow was so strong and the person’s location was lost. Within that first couple of minutes I came up with a system that I thought might be able to help.”
There are an average of 135 drownings in Ireland a year, and Deepseek’s technology could potentially reduce that number significantly.
Mulcahy founded the company alongside Ciaran O’Meara during a final year engineering project in University of Limerick.
“I did an undergraduate in electronic and computer engineering in University of Limerick. I just graduated from that this year, so I’m just 22 years old. I’m starting a PhD in researching AI for search and rescue in the University of Limerick now.
“At the moment we’ve just had the €7,000 investment from the Ireland’s Best Young Entrepreneur programme and we haven’t required any more funding at the moment – I suppose we are privileged in a way that we have the technology knowledge within the company, with the founding members, so we are able to develop the technology ourselves and scale from there.
“It’s just using an autonomous drone-based system that uses an AI system in its camera, so you take the feed from the camera and you run it against an AI algorithm and it checks if there’s someone in the water. It uses thermal and visual light and then it essentially automates the search component out of the search and rescue so when the rescue teams arrive in their boats they know exactly where to go and they can make the rescue a lot more efficiently.”
There are already drone systems in use within the rescue services in Ireland, and Deepseek is focusing on cutting out the human operating the drone. It will work more in tandem like a co-worker rather than a piece equipment.
“We are trying to automate the whole drone system so there is no human interaction – its completely autonomous – that’s where the AI comes in. It’s essentially putting eyes on the drone and the drone is able to figure out for itself if there is someone in an image and then feed that information back because the real critical thing in a river rescue is the time, that initial couple of minutes is incredibly important in a river rescue because things happen at such a rapid pace and so the drone needs to be able to think for itself and just feed back the information that’s needed.
“We’d need an amount of co-operation from everyone to try and develop a system that fits for everyone and that’s able to provide them with the information they need so that’s what we are working on currently.
“We are talking with the Limerick Search and Rescue cohort, the coast guard and the main search and rescue. We’ve been in with them in their base in Limerick and we are just trying to figure out at the moment how to integrate it into their system, that’s really our plan.”
While Deepseek is concentrating only on river rescues until it gets its initial AI software off the ground, the potential this technology has to be used in other areas is huge. It currently takes about 15 minutes for helicopters to be off the ground after receiving a call during daylight and 45 minutes at night – an unmanned drone could have a much faster response rate.
“We are just a software company so we are not going to be developing any specific drone hardware, we buy that off the shelf and then we extend the off the shelf drone products with our own software so we wouldn’t, essentially, be manufacturing drones or anything like that which makes things easier on our part, and makes things a lot leaner as well. We can move around and put our software in different types of drones, we won’t be tied in.”