This innovative swarm approach created for drone navigation, enhances speed and efficiency with advanced AI technology.
Researchers at Shanghai Jiao Tong University have developed a lightweight, artificial intelligence based system that enables drone swarms to navigate complex environments at high speeds without relying on expensive hardware or human control.
The method represents a major advance in swarm robotics and could be applied in disaster response, inspection, autonomous filming, and other practical fields.Most current drone navigation systems depend on a modular structure where tasks such as mapping, path planning, obstacle detection, and flight control are handled separately.While effective in some cases, this structure increases the risk of accumulated errors and slow reaction times, especially in crowded or fast-changing environments.The team in Shanghai replaced this traditional structure with a compact, end-to-end neural network using differentiable physics.Swarm intelligence redefinedThis method allows the system to learn flight control directly by simulating physics and adjusting itself through what is called “backpropagation.” This dramatically improves learning speed and real-world performance.Also, instead of using high-resolution cameras or expensive sensors, the system uses an ultra-low-resolution 12 by 16 pixel depth camera. This low-fidelity input, the team explains, is similar to the compound eyes of insects.The network uses this limited data to make real-time decisions and steer the drone through obstacles and confined spaces. In testing, drones equipped with the system flew at speeds of up to 65.6 feet per second. This is twice as fast as earlier learning-based methods.The system also achieved a 90 percent success rate in navigating cluttered spaces, compared to about 60 percent in previous studies. One major benefit is that the neural network runs on a small development board costing only $21.It performs well without needing a graphics processing unit, making it possible to build and deploy many drones at low cost. This could significantly lower the barrier to entry for swarm-based drone operations.Unlike many deep learning models that require large, expert-labeled datasets, this system was trained entirely in simulation. The team used basic geometric environments and embedded physics to teach the model how to fly. Trained on simulation not dataInterestingly, no real-world flight logs or handcrafted training data were used. In swarm tests, multiple drones were able to coordinate their movements without communicating with each other or relying on a central controller.This self-organized behavior is uncommon in existing robotic swarm systems and shows strong potential for future applications. The research group is now working on replacing the depth camera with optical flow sensors, which track motion and are commonly used in insect vision.They are also investigating how the AI model makes decisions, in order to better understand its internal reasoning. “We are currently exploring the use of optical flow instead of depth maps for fully autonomous flight,” explained Prof. Zou and Prof. Lin. “Optical flow provides fundamental motion cues and has long been studied in neuroscience as a key component of insect vision,” he added. “By using it, we hope to get even closer to mimicking the natural strategies that insects use for navigation. Another important direction we’re pursuing is the interpretability of end-to-end learning systems,” he explainedReaders can view the complete study in the journal Nature Machine Intelligence or through the arXiv open-access archive.
Military Drone Drone Swarms Energy &Amp Environment Inventions And Machines UAV
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Virus Discovery Among Bats in China Fruit Orchards Draw Exposure ConcernsThe Best in Science News and Amazing Breakthroughs
Read more »
President Trump’s Iran strategy is bearing fruitIran is still dangerous, but the world is safer today with its weaknesses exposed.
Read more »
This Underrated Fruit Is Packed With Vitamin C (And A+ For Immune Health)Brew it as a tea for a sweet-tart treat.
Read more »
How to Get Rid of Fruit Flies and More, According to Pest Control ExpertsCory Fernandez is a commerce editor for NBC Select, and an authority on all things food, cooking, kitchen, home, and general lifestyle.
Read more »
Tropical fruit now accepted as currency at these Florida restaurantsAs mangoes reach peak ripeness, some Miami-area restaurants are accepting them as currency, offering cocktails, desserts and bread in exchange for the tropical fruit.
Read more »
Sing along to ‘Fruit Fly,’ homegrown S.F. musical at the Roxie next weekThe Mission-based filmmaker hosts a sing-along party for his musical ode to San Francisco's LGBTQ community at the Roxie Theater on July 23.
Read more »