Carnegie Mellon’s Autonomous Drones Master Obstacle Navigation with Real-Time 3D Mapping
Researchers at Carnegie Mellon University have come up with some futuristic tech to help drones autonomously navigate around obstacles. No more crashing into walls or people for these flying bots!
The team, led by Professor Kenji Shimada, tested their system at a live construction site in Japan. The goal was to see if drones could zip through the tunnel while avoiding workers and equipment. The drones totally crushed it!
Shimada said companies want to use drones for dangerous jobs so people don’t have to risk their lives anymore. But first, drones need to be able to handle complex, changing environments – not just open skies.
That’s why the team developed a way for drones to build 3D maps of their surroundings in real-time. This lets them track where stuff is and anticipate where people and objects are headed.
The drones used this tech to scan the shape of the tunnel and identify finished sections versus areas still under construction. This helped guide the building process without putting workers in harm’s way.
To predict moving obstacles like people, the drones used a hybrid 3D map to monitor both stationary and dynamic elements. An algorithm then determined potential collision points and rerouted the drone to avoid mishaps.
By refining their obstacle avoidance capabilities, the researchers enabled the drones to navigate smoothly through the active construction site. No crashes or close calls!
Shimada called it the first time autonomous 3D scanning drones have been used in a setting like this. And it worked flawlessly! The team believes their tech will make worksites much safer by letting drones take over risky inspection tasks.
Carnegie Mellon has developed some seriously smart navigation tech for drones. These flying bots can now autonomously zip around obstacles in complex, changing environments. No more bonking into walls or people!