Developing AI Drones that Fly Autonomously to Perform Inspections and Other Tasks
We are developing technology that allows drones to fly autonomously based on image information from a camera mounted on the drone.
By utilizing this technology, we aim to significantly improve the efficiency of tasks that have traditionally been performed by humans, such as patrol inspections and inventory management.
Automomous Drone Application Examples
Example #1: Automated patrol in factories and plants
Autonomous drone flights can be used to capture images of the work site, replacing daily human patrols and inspections. Simply draw a line and a commercially available drone will autonomously patrol the indoor area.
- Easy to install in the field
-- Just put a line tape on the flight route.
-- No need to create a map in advance.
Demonstration: Automatic patrol in office
Example #2: Inventory in warehouse
The drone flies autonomously to trace all products, scans the barcodes and QR codes affixed to the inventory, and registers the inventory in the inventory management system. You can expect reduced need for labor, improvement of inventory accuracy, shortening of operation suspension period, etc.
Example #3: Power line inspection
The drone flies along a flexed power line and captures images / videos of the power line.
At that time, by controlling the transmission line so that it is kept at a certain distance, it is possible to always take a picture of the transmission line at a resolution appropriate for inspection.
Deep Learning Technology for Autonomous Flight
Deep learning is applied to drone control, and the direction of movement of the aircraft is autonomously determined from the current camera image. Specifically, the drone is trained using a deep neural network that inputs image data and outputs drone control commands (left turn, straight ahead, right turn)known as supervised learning or reinforcement learning.