It highlights the need for AI to learn and adapt to the real world through interaction and training. Overall, the success of the Swift AI in drone racing demonstrates the potential for AI systems to excel in physical sports that are less predictable than board or video games. Additionally, fast autonomous drones could be used for shooting action scenes in the film industry, and the speed of these drones could be crucial for rescue missions in buildings on fire. The ability to fly at high speeds could be valuable for covering large areas in a limited time, such as in forest monitoring or space exploration. The researchers believe that Swift’s capabilities could have real-world applications in environmental monitoring, disaster response, and the film industry. However, human pilots showed more adaptability than the autonomous drone, performing better in conditions different from what Swift was trained for. But if you’re looking for an alarm, remote starter, along with the DroneMobile module and installation, you’re looking at a starting price of 400, depending on your vehicle’s type. Swift achieved the fastest lap, surpassing the best lap by a human pilot by half a second. The races took place on a purpose-built track in an airport hangar near Zurich. Small drones are being utilized in monitoring, transport, safety and disaster management, and other domains. This training method helped to avoid the destruction of multiple drones during the learning process.Īfter a month of simulated flight time, Swift was ready to compete against human pilots, including world champions. Through trial and error, Swift taught itself to fly and learned to autocorrect errors it made while interpreting data from the onboard sensors. Connects your Compustar remote start or security system to your smartphone. Instead, it was trained in a simulated environment using reinforcement learning. Unlike most unmanned drone racers, Swift does not rely on an external position-tracking system to control its trajectory. This information is then fed to a control unit, which uses a deep neural network to choose the best action for completing the circuit as quickly as possible. It uses an integrated inertial measurement unit to measure acceleration and speed, and an artificial neural network processes the camera data to determine the drone’s location in space and detect gates along the racetrack. The Swift AI pilot reacts in real time to data collected by the onboard camera, similar to human racers. The AI, called Swift, participated in first-person view (FPV) drone races where pilots control quadcopters at speeds exceeding 100 km/h while wearing headsets linked to onboard cameras. At this time, this issue is only affecting DroneMobile systems with a serial number starting with '20'. Researchers at the University of Zurich have developed an AI system that has successfully beaten world champions in drone racing.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |