University of Canterbury computer science professor Richard Green (left), and UAV expert Dr Sam Schofield with their chainsaw drone. Photo / Sacha Skinner, University of Canterbury
University of Canterbury computer science professor Richard Green (left), and UAV expert Dr Sam Schofield with their chainsaw drone. Photo / Sacha Skinner, University of Canterbury
Chainsaws, drones and AI might not scream safety – but a University of Canterbury team has built a chainsaw drone that could revolutionise high-risk arborist work.
Leader of the project, computer science professor Richard Green, said using automation had progressed extensively over the years, but it had reached a pointwhere it needed computers that understood the 3D environment they moved in and could actually interact with it.
“Over the last eight years, we’ve been developing unmanned aerial vehicles [UAV] like drones,” he said.
“Before submitting our proposal, we met with lots of different industries to see what could be useful to explore.
“We didn’t want to be a group of academics coming up with solutions that may not actually be needed.”
The project has received a $10 million grant over five years from the Ministry of Business, Innovation and Employment.
The team includes Green, UC mechanical engineering professor Dan Zhao, UAV expert Dr Sam Schofield and University of Auckland mechanical engineering professor Karl Stol.
One topic was the safety of trimming trees around live powerlines while nearby arborist work was carried out.
“Initially, we developed a drone to carry an electric pruning tool that trimmed smaller branches, offering really good precision.”
A chainsaw drone at work using AI software to cut small branches off trees. Photo / Sacha Skinner, University of Canterbury
For branches larger than 30mm in diameter, a chainsaw drone was developed.
“We’re using a small battery-driven chainsaw that can prune branches and twigs larger than 30 millimetres in diameter.
“We use a DJI Matrice drone, which has a USB-C socket interface for our processor.
“When we plug the processor in, we can see all the sensors and fly it using the controller.
Adjusting the chainsaw drone during trials. Photo / Sacha Skinner, University of Canterbury
“We fly near to a branch, tap the branch on the screen, and the drone takes over, using AI technology.”
The drone operators have been working with drones for 10 years and are very experienced flyers.
The AI in the computer processor is “trained” using thousands of examples of branches, negotiating leaves, twigs and wind and learning how to auto-navigate once a branch has been identified.
Green said the team, along with Kiwi and international UAV experts, researchers and manufacturers, hoped to produce AI-driven chainsaw drones commercially next year for use across a range of different industries where jobs are inaccessible and hazardous for humans.
