Breakthrough in combatting Eurasian watermilfoil,artificial intelligence-powered robot supports waterfront residents
By William Crooks
Local Journalism Initiative
A breakthrough has been reached in the fight against Eurasian watermilfoil, an invasive plant species plaguing local bodies of water. Bishop’s University (BU), in partnership with RAPPEL (Regroupement des associations pour la protection de l’environnement des lacs et des bassins versants), has put forth the world’s first application of an aquatic drone, powered by artificial intelligence (AI), to improve the efficiency of efforts to control against invasive aquatic plants. The Record attended a press conference in Austin on Aug. 29 for more information.
The project, led by the Undergraduate Bishop’s Earth Research Group (UBERG), specifically Steven Poulin, a master’s student in Computer Sciences at BU, involved an aquatic drone (UBER-Gaiter) criss-crossing lakes in the Estrie and Outaouais regions over the past year. Its goal was to build a bank of images for its deep-learning AI system. Over 60,000 images of aquatic plants have been integrated within the drone, which can now autonomously recognize twelve aquatic plant species with over 95 per cent efficiency, surpassing human performance in this area.
RAPPEL, a group of strategy experts in the control of Eurasian watermilfoil, proceeded, in recent weeks, to the concrete application of the drone for its own operations. In collaboration with the Association pour la protection de l’environnement du lac O’Malley (APELO), RAPPEL used the aquatic drone in Austin at O’Malley Lake. The drone first surveyed the waters to collect data. RAPPEL then used the data, the mapping and identification of Eurasian watermilfoil, for its own work. This resulted in operations for the manual removal of the plants based on precision “geofencing”: a world’s first.
This scientific advance is significant because it will reduce the cost of controlling invasive exotic aquatic plants. The mapping produced using the drone provides an accuracy of 10 to 15 cm, reducing the positioning time of the divers who carry out the uprooting of the plants underwater. The future applications of such technology to improve knowledge of an aquatic environment and to carry out floristic inventories are also very promising.
The research team now hopes to obtain the necessary funds to produce other drones of the same type and benefit a greater proportion of water protection stakeholders.