How AI Could Help You Control Your Next Drone in Tough Conditions

Drones are getting smarter and easier to control thanks to artificial intelligence (AI). 

Researchers have devised a new machine-learning approach that could be used to control a robot, such as a drone or autonomous vehicle, more effectively and efficiently in environments where conditions change rapidly. It's part of a growing wave of AI innovations that could transform how drones are used for hobbyists, the military, and industrial applications. 

"AI reduces the need for humans to be constantly in-the-loop," Carlo Pinciroli, a professor of robotics engineering at Worcester Polytechnic Institute, told Lifewire in an email interview. "This will allow operators to treat drones as teammates rather than 'puppets,' making complex and dangerous missions possible while minimizing the risk of human harm." 

Drones Flying Smarter

The new technique MIT and Stanford University scientists created could let a drone closely follow a downhill skier despite being buffeted by strong winds. The researchers have integrated elements of control theory into their learning process, a kind of valuable hint that guides the system's control strategy.

The researchers found their method was data-efficient, so it could effectively model a highly dynamic rotor-driven vehicle using only 100 data points. This efficiency could make their technique especially useful when a drone or robot needs to learn quickly in rapidly changing conditions.

"Our approach is inspired by how roboticists use physics to derive simpler models for robots," Spencer M. Richards, the lead author of a paper on the new technique, said in a news release. 

The autonomy provided by AI allows drones to perform complex tasks such as mapping large areas or conducting search-and-rescue operations, David Ly, the CEO of Iveda, which specializes in AI-powered smart drones, said via email. AI-powered drones can adapt to changing environments and handle unexpected situations.

"Moreover, AI algorithms can analyze data from various sensors on the drone in real time, enhancing the drone's capabilities for data collection and analysis," he added. 

The film and photography industry has embraced the capabilities of AI-powered drones to get spectacular shots of live events, movie scenes, and natural landscapes, Pinciroli said. Drones are being developed for last-mile delivery service of small packages.

"In this case, AI enables the drones to both navigate efficiently to their destination and to identify a safe landing spot to deliver the parcel autonomously," he added. 

Drones can also be used to conduct remote visual monitoring and inspections of aging infrastructure, Dustin Parkman, vice president of Transportation Solutions at Bentley Systems, said via email. One example is using a drone to monitor and inspect bridges for cracks and defects. The drone imagery can be used to make digital twins by transforming drone imagery into 3D bridge models. 

"The drones can be sent on autonomous flight paths to get the right shots," he added. "You can give a drone a path and have it capture the right quantity and quality of data for a proper inspection."

The Future of AI-Powered Drones 

AI is also transforming the way drones are used in warfare. The conflict in Ukraine is driving rapid advances in how AI controls drones. 

"Modern warfare is highly adaptive in its nature, and AI, if done right, can adapt to rapidly changing environments and threats," Adam Robertson, CTO of Fortem Technologies, an airspace awareness company, said in an email to Lifewire. "AI allows for the automation of many things that historically required a human, such as a human on the lookout for incoming aircraft, which is now automated with AI and image recognition.

AI can also be used to augment human abilities, Robertson said. Fortem's DroneHunter, for example, uses AI to handle the pursuit and destruction of enemy drones. A human gives the gadget permission to pursue and permission to attack. 

"This high-level automation speeds up the mission, eliminates threats faster, cuts human error to near zero, reduces the operator skill required, simplifies the operator workload, and increases the success rate dramatically," he added. 

In the future, large AI language models like ChatGPT may be used to control drones by voice.  

"A verbal command from an officer with authority could be used for voice identification/authorization and for a command such as 'maintain a 24/7 search for drone threats,'" Robertson said. "Once a command is made, AI will take care of the details from there."