A drone can guide itself and map environments via echolocation using a simple buzzer and microphone set-up, much like how a bat uses sound to see in the dark.
For robots to be able to move autonomously, they need to determine where they are in space and whether any obstacles lie in their path. This is typically done using GPS sensors, cameras or devices that use radio waves or lasers, but these can be too heavy for small robots and might fail in certain environments, such as when it is dark or there is a lot of smoke.
Now, at the University of Toronto in Canada and her colleagues have developed a system that can map the walls of a room to an accuracy of up to 2 centimetres using a simple microphone and speaker.
Advertisement
They tried this approach because a lot of robots have microphones already, says D眉mbgen, who did the work while at the Swiss Federal Institute of Technology in Lausanne, Switzerland. 鈥淚f the robot is designed to interact with humans, for instance, it would have microphones to hear commands,鈥� she says.
To work, the speaker emits bursts of sound containing a range of frequencies, which bounce off walls and are recorded by the microphone when they return. An algorithm then reconstructs the surfaces of the environment based on how the sound wave has interfered with its reflections.
D眉mbgen and her team tested the system on a small drone, on which they installed a buzzer and four microphones, and on a wheeled robot about the size of a tennis ball, which already had a small speaker and microphone.
When the robots were still, they could map walls with a precision of 2 centimetres from half a metre away, although this increased to 8 centimetres when the drone was flying.
While these figures are less accurate than the best laser and camera-based systems, it could serve as a useful alternative for when cost or weight is a factor, or in difficult-to-map environments, says D眉mbgen.
A robot using the technique can also make a map of its environment and localise itself at the same time, which means it can return to its start position or follow a certain path through the room without previous knowledge of the environment, an essential tool for search-and-rescue tasks.
鈥淒oing localisation with sound is quite a beautiful solution because of the small scale and the low cost and availability of these microphones,鈥� says at the University of Seville in Spain. While it demonstrates a useful level of accuracy, the team will also need to show that it can work in more complex, real-world environments, says Zufferey.
Reference:
