Nature has always served as a source of inspiration for scientists and roboticists seeking to create new characteristics for machines. In this example, the University of Toronto researchers were inspired by bats and other creatures that use echolocation to develop a mechanism that would allow small robots to navigate themselves — one that does not need costly gear or components that are too bulky or heavy for tiny machines. According to PopSci, the team merely utilised the inbuilt audio hardware of an interactive puck robot and developed an audio extension deck out of a cheap microphone and speakers for a little flying drone that fits in the palm of your hand.
The mechanism functions similarly to bat echolocation. It was created to generate noises of various frequencies, which a robot’s microphone takes up when they bounce off walls. The team’s algorithm then gets to work analysing sound waves and creating a map of the room’s dimensions.
The researchers said in an article published in IEEE Robotics and Automation Letters that current “algorithms for active echolocation are less developed and sometimes depend on hardware requirements that are out of reach for tiny robots.” Their “method is model-based, operates in real-time, and needs no previous calibration or training,” they said. Their technique might enable tiny machines to be dispatched on search-and-rescue operations or to previously unexplored places that larger robots could not access. And, since the system simply requires onboard audio or inexpensive extra hardware, it offers a broad variety of possible applications.
During their experiments, the researchers discovered that their approach is still not as precise as systems that require larger and more costly gear, such as GPS sensors or webcams. In future iterations, they want to increase its accuracy and remove the requirement for the system to make noises. Instead, they want their system to be able to echolocate using noises generated by the drone itself, such as the spin of its rotors.
