Underwater drone vacuums up lost golf balls from lakes

With a “woosh”, golf balls zoom across the course towards the green, but every once in a while, a lake comes in the way, and a ball is lost.

The golfer takes out a new ball and continues the game while the lost ball settles into the lake.

It can then lie there for up to a year before a diver is sent out to find it, says Haroon Sajjad, master’s student in Technology Entrepreneurship at DTU.

“A golf ball can dissolve quite a bit in a year, which both causes pollution and makes it less possible to reuse the ball. So we would like to offer a solution where, based on a subscription, we come by once a month with an underwater drone and pick up the balls. Three clubs have already shown interest,” he says.

Haroon Sajjad works together with four—soon to be six—fellow students in their startup company ODrone, which is short for “Ocean Drone”. The idea is, in the long term, to use the drone to collect waste in the oceans.

Initially, however, the drone will search for golf balls, and the group is well underway with development as one of 19 selected projects in the accelerator programme Danish Tech Challenge, which is taking place at the DTU Science Park over the next few months.

“It started with us having a course together at DTU called ‘Digital Trends’, where we talked about which technologies could be used to solve concrete problems,” Haroon Sajjad says.

“At the same time, I read an article about someone who had made a lot of money as a child collecting golf balls to resell them. At least until the golf ball retailers got mad that he was ruining their business. Today, however, there is much more focus on the environment, so now it makes sense for everyone to collect them,” he says.

Millions of golf balls go missing every year

A PhD thesis from Stanford University has shown that golf balls cause a lot of pollution over time, as both varnish and colour on the surface, as well as parts of the core that contains plastic, are released in the water.

The researchers had collected almost 40,000 balls from the coasts of California near five golf courses and observed how the cellular plastic material polyurethane had continuously dissolved from the surface.

They estimated that just under 30 kg of micro- and nano-plastics had been discharged from the balls into the sea before they were collected.

“So with 600 million missing golf balls a year across the world, it adds up to quite a lot,” Haroon Sajjad says.

ODrone has had the parts for the drone delivered through DTU and has then begun to modify it from top to bottom.

In order to be able to move around the lake in all three dimensions, the drone is equipped with six motors, and it orients itself using both a front camera and a GoPro camera, which provides a better view of the murky water than the fixed camera.

“It has been a challenge for us to get the drone to move properly together with the robot arm, as all the parts have to move in a super coordinated way, so that the drone can swim straight,” Haroon Sajjad points out.

The robot arm will be scrapped

The algorithm, which makes the decision about what is a golf ball and what should happen to it, was developed by the group themselves in Python, and the electronics are mounted on a Raspberry Pi, which is placed in a waterproof chamber in the centre of the drone.

It has been trained on a large number of images and videos of golf balls in more or less turbid lakes and through tests in a swimming pool. In particular, it is the golf balls’ colours, sizes, and the familiar depressions in the balls that the cameras look for, and the drone will continuously refine its skills during use.

During the tests, the drone was equipped with a robotic arm that picked up the balls with a “mouth” that could open and close around the ball to lead it into a box. But it proved difficult to control its motor.

“Our vision technology has been able to recognize 95 percent of the balls, and we have so far picked up 10-15 of them with the robotic arm. But after two or three tests in a pool, water got into the arm’s electronics, so it became too complex to work with,” Haroon Sajjad says.

Therefore, the plan is now to mount a kind of vacuum cleaner at the bottom of the drone, which instead sucks the balls up into a chamber.

“The drone will function somewhat like a robotic vacuum cleaner that scans the lake, lays out a route, and starts sucking up the golf balls,” Haroon Sajjad says.

There are also thoughts about several cameras and a sonar sensor and possibly even a radar sensor to give the drone the best imaginable opportunities to orientate itself in both deep and shallow water as well as steer around both land and algae without stopping.

In talks with a golf club

According to Haroon Sajjad, the group is in talks with Furesø Golfklub, so that the drone can be tested in the right environment, and then the plan is to move away from remote control to make the drone work more autonomously.

As for the weight, there is also a lot of work to be done. It weighs around 10 kg, so it sinks as soon as it enters water.

The group is also looking at materials such as carbon fibre in order to make it float, so that it would in the future be able to surround floating plastic or oil spills on the surface of rivers and seas.

“If there is plastic or oil floating on the surface of the water, the drone must be able to float to it and catch it with a net like a lasso,” Haroon Sajjad says.

The battery therefore also becomes a factor, as it currently only lasts for four hours, and a mission at sea is expected to drain it faster due to, among other things, the stronger underwater currents. At the same time, algorithms and cameras must be upgraded to be able to tell apart many different types of materials if the drone is to be used at sea.

“So we’re starting out with the golf balls, but we don’t lack ideas and ambitions,” Haroon Sajjad says.

(This article was written together with our 9th grade intern Signe Bredskov.)