Autonomous Rescue Drones Navigate Forest Trails Using Artificial Intelligence Algorithm
A team of Swiss researchers have succeeded in teaching a drone how to spot and follow trails in forests. The quadcopter used by the team is equipped with two cameras and a powerful artificial intelligence algorithm that processes the images and identifies man-made trails. During tests, the drone was able to successfully identify 85% of the trails while a human could identify only 82%. These drones could be used in search and rescue operations in dense forests.
There are similar researches of <a href="https://www.crazyengineers.com/threads/mit-demos-route-planning-algorithm-drones-avoid-obstacles-on-the-fly.86499">MIT Demos Route Planning Algorithm - Drones Avoid Obstacles On The Fly</a> based on acoustics and sensors, but this is the first venture into artificial intelligence algorithms. The researchers call it Deep Neural Network (DNN), which is similar to the functioning of a human brain. The algorithm is capable of learning new scenarios and considering them in future calculations.
First, the team had hiked in the Alps wearing helmets with 3 adventure cameras to take more than 20000 images of trails ahead, to the left and right. These images were then used to train the algorithm to identify the trails. This led to successful identification of previously unseen trails almost 85% of the times. Researchers say that an image of a forest is very difficult even for a human to decode and the algorithm, though in the preliminary stage, is doing a great job.
Dalle Molle Institute for Artificial Intelligence, University of Zurich and NCCR Robotics together have pioneered this trail-identifying DNN algorithm, and say the next step is to teach it human-detection. In Switzerland alone, there are about 1000 cases annually of lost hikers calling in for help, and this intelligent drone could be a boon. Though there is a lot to be done before autonomous fleets of drones can be deployed alongside humans in rescue operations, they will be more efficient and cost-effective than human rescuers alone.
Source: #-Link-Snipped-# | <a href="https://robohub.org/drones-recognise-and-follow-forest-trails-in-search-of-lost-people/" target="_blank" rel="nofollow noopener noreferrer">Drones recognise and follow forest trails in search of lost people - Robohub</a>
There are similar researches of <a href="https://www.crazyengineers.com/threads/mit-demos-route-planning-algorithm-drones-avoid-obstacles-on-the-fly.86499">MIT Demos Route Planning Algorithm - Drones Avoid Obstacles On The Fly</a> based on acoustics and sensors, but this is the first venture into artificial intelligence algorithms. The researchers call it Deep Neural Network (DNN), which is similar to the functioning of a human brain. The algorithm is capable of learning new scenarios and considering them in future calculations.
First, the team had hiked in the Alps wearing helmets with 3 adventure cameras to take more than 20000 images of trails ahead, to the left and right. These images were then used to train the algorithm to identify the trails. This led to successful identification of previously unseen trails almost 85% of the times. Researchers say that an image of a forest is very difficult even for a human to decode and the algorithm, though in the preliminary stage, is doing a great job.
Dalle Molle Institute for Artificial Intelligence, University of Zurich and NCCR Robotics together have pioneered this trail-identifying DNN algorithm, and say the next step is to teach it human-detection. In Switzerland alone, there are about 1000 cases annually of lost hikers calling in for help, and this intelligent drone could be a boon. Though there is a lot to be done before autonomous fleets of drones can be deployed alongside humans in rescue operations, they will be more efficient and cost-effective than human rescuers alone.
Source: #-Link-Snipped-# | <a href="https://robohub.org/drones-recognise-and-follow-forest-trails-in-search-of-lost-people/" target="_blank" rel="nofollow noopener noreferrer">Drones recognise and follow forest trails in search of lost people - Robohub</a>
0