Carding 4 Carders
Professional
Robots are now on the same page as the sounds around them.
For safe co-existence with humans, robots must be able to detect their presence and determine their location to avoid accidents and collisions. Until now, most robots have been trained to locate people using computer vision techniques based on cameras or other visual sensors.
A research team from the Georgia Institute of Technology has developed an alternative method for human localization that relies on subtle sounds that naturally occur when moving in a particular environment. This method can be applied to a wide range of robotic systems.
The acoustic localization method proposed by the researchers is based on machine learning algorithms. To train their algorithms, the team created a special data set that contains 14 hours of high-quality audio recordings, compared with video recordings.
The machine learning technique developed by the researchers is trained to localize people solely based on sound. Since it requires only audio recordings obtained with microphones, it can theoretically be implemented in any robot with an integrated microphone.
The researchers trained their model to ignore external and unrelated noise. In initial tests, they tested their technique on the Stretch RE-1 robot, a compact robotic arm.
In initial tests with the Stretch RE-1 robot, the team's technique performed twice as well as other acoustic localization techniques. These results highlight the feasibility of acoustic localization, which is highly scalable and less intrusive than camera-based localization.
In the future, the human localization technique developed by the researchers may help improve the security and performance of robots designed to work closely with humans, while maintaining the privacy of users.
For safe co-existence with humans, robots must be able to detect their presence and determine their location to avoid accidents and collisions. Until now, most robots have been trained to locate people using computer vision techniques based on cameras or other visual sensors.
A research team from the Georgia Institute of Technology has developed an alternative method for human localization that relies on subtle sounds that naturally occur when moving in a particular environment. This method can be applied to a wide range of robotic systems.
The acoustic localization method proposed by the researchers is based on machine learning algorithms. To train their algorithms, the team created a special data set that contains 14 hours of high-quality audio recordings, compared with video recordings.
The machine learning technique developed by the researchers is trained to localize people solely based on sound. Since it requires only audio recordings obtained with microphones, it can theoretically be implemented in any robot with an integrated microphone.
The researchers trained their model to ignore external and unrelated noise. In initial tests, they tested their technique on the Stretch RE-1 robot, a compact robotic arm.
In initial tests with the Stretch RE-1 robot, the team's technique performed twice as well as other acoustic localization techniques. These results highlight the feasibility of acoustic localization, which is highly scalable and less intrusive than camera-based localization.
In the future, the human localization technique developed by the researchers may help improve the security and performance of robots designed to work closely with humans, while maintaining the privacy of users.
