Micro Drones Equipped with Optical Sensors Has Been Developed that Can Explore Unknown Environments Completely Autonomously
Posted 09 Dec-2019 02:01 PM by Rachel |
According to foreign media reports, researchers at the Delft University of Technology have demonstrated a group of miniature drones equipped with optical sensors that can explore unknown environments completely autonomously. This micro drone is inspired by the swarm of insects. It relies on many inexpensive robots with relatively limited sensing and computing technology. Compared to a single large drone, these robots can explore larger areas with greater speed and efficiency.
The biggest challenge in achieving swarm detection is the level of personal intelligence of drones," said Kimberly McGuire, a doctoral student involved in the project.
Researchers first studied the basic flying capabilities of drones, and then devised a method to ensure that the drones would not hit each other. For this reason, researchers have each drone carry a wireless communication chip, and then use the signal strength between the chips to avoid collisions. This method requires very little calculation. Guide de Croon, the main investigator of the project, explained: "Because they are very small, they are equipped with only the smallest sensors." According to researchers, drones have inertial measurement units consisting of accelerometers and gyroscopes, optical flow computer mouse sensors, and lasers. Among them, there are lasers pointing down to measure the horizontal speed of the drone, and four other lasers measuring the point distances on all sides of the drone. The researchers explained: "The laser is set in a cross shape, pointing forward, rear, left and right. We have demonstrated that with these sensors, drones can explore unknown environments on their own."
At the same time, the researchers conducted an experimental project in which the drone was equipped with a camera and deployed in an indoor office environment of about 40 × 12 meters. There are two dummies in the office representing the victims in the disaster scenario. "In an experiment, 'rescue' personnel could look at these images to see if a 'victim' exists in the environment and where the victim is," the researchers said.
In six minutes, a group of six drones can detect about 80% of the space in the building, which is impossible for a drone. In addition, if a drone finds a victim, but cannot bring back the image due to a hardware failure, another drone will capture the victim's image and quickly return it to rescue personnel.
The researchers said, "The main idea of the new navigation method is to reduce our navigation expectations to the extreme: we only require the robot to be able to navigate back to the base station. By having each robot follow a different preferred direction before letting them spread into the environment. After the exploration, the robot will return to the wireless beacon located at the base station. "
Detailed maps are very convenient in principle as they allow the robot to navigate the best path from any point on the map to other points. However, the cost of making such a map on a miniature robot is too high. Although the proposed wrong algorithm leads to low path efficiency, its advantage is that it can achieve the same purpose on micro robots.
ccording to researchers, in the future, drones can detect the victims through airborne processing and stay there to make a loud tweet, or they can signal other drones in the group. And a chain from the victim to the base station is formed, and rescuers can follow the chain.