Imagine if drones could evolve like animals, where a tiny mosquito and a giant eagle have different bodies for different jobs. Instead of humans guessing the best shape, AI 'breeds' the perfect robot body and brain together. This means the drone's physical build and its software are tailor-made for the specific environment it needs to fly in.
Current drone design is limited to a few standard shapes, like quadrotors, regardless of the mission. This leads to inefficiency and high manual engineering time when a specific task requires a non-standard robot shape.
An AI-driven system that uses evolutionary algorithms and deep learning to automatically design the physical body and control software of aerial robots.
If you are a delivery service dealing with diverse urban environments — this project developed a method to create task-optimized flying machines that are more efficient and resilient. This allows for drones that fit the specific mission rather than using a one-size-fits-all quadrotor.
If you are a rescue operator dealing with dangerous or tight spaces — this project developed AI-designed robots using soft and rigid materials. This results in intrinsically safe flying robots that can navigate complex areas more effectively than standard drones.
If you are an inspection firm dealing with time-consuming custom drone builds for different sites — this project developed an automated design process. This replaces manual engineering with AI that creates the best fit for the environment, reducing design time.
Based on available project data, there is no specific pricing or cost information provided.
The project involves 2 industry partners and 3 SMEs, suggesting a move toward industrial application, though specific scaling metrics are not listed.
Based on available project data, the licensing and IP terms are not specified in the provided summary.
The project period runs from 2023-11-01 to 2027-10-31, indicating the research phase concludes in late 2027.
The project uses deep neural networks for navigation strategy learning to create a tight link between the robot's physical body and its brain.
The consortium is heavily weighted toward academia with 6 universities and 2 research institutes, reflecting the high-risk, high-reward nature of the research. However, the inclusion of 2 industry partners and 3 SMEs across 7 countries provides a necessary bridge to commercial application, ensuring that the AI-driven designs are grounded in industrial reality.
Contact NORGES TEKNISK-NATURVITENSKAPELIGE UNIVERSITET NTNU
Contact us to track the evolution of this AI-robotics design tool.
