Sonicsense: the progress of the robotic touch | Bcg generative ai pdf | Generative ai examples in healthcare | Examples of generative ai tools | Turtles AI

Duke University researchers created Sonicsense, a system that allows robots to recognize materials and shapes through vibrations. Thanks to integrated microphones, the device offers a new approach to robotic perception. Sonicsense represents a step forward in improving the robot-environment interaction.

Key points:

• Sonicsense provides robots with a new method of perception based on vibrations.
• The system consists of a robotic hand equipped with contact microphones.
• It can identify materials and shapes with a reduced number of interactions.
• Construction costs are contained thanks to the use of commercial components.

Sonicsense is an advanced system developed at Duke University, designed to equip the robots with a "tactile" ability through the analysis of vibrations generated in contact with different objects. This technology is based on a four -finger robotic hand, in which each tip is equipped with a contact microphone. These microphones are able to capture the vibrations that are produced during the interactions, excluding external noises and allowing the robot to focus on the characteristics of the manipulated object. The idea is to offer robots a deeper and complex perception, similar to the human one, to improve their ability to understand the surrounding world. The researchers, led by PhD student Jiaxun Liu and Professor Boyuan Chen, stressed that although vision is fundamental for robots, the sound analysis offers an additional information level, revealing details that the view may not capture. Through the interpretation of the signs collected, Sonicsense is able to extract frequency characteristics, thus allowing to identify the material of an object and its three -dimensional form. In situations in which the object is unknown, the system can require up to twenty interactions to provide careful identification, while for already known objects a significantly lower number of interactions is sufficient. Among the features demonstrated by Sonicsense, there is the ability to count objects within a container or determine the level of liquid in a bottle, all by exploiting the vibrations generated by the movement. The system is designed to operate in open laboratory environments, replicating real interactions rather than controlled conditions, which represents an important challenge in the field of robotics. The low cost of about $ 200 for the construction of the device is made possible by the use of commercial microphones and 3D printing. The research team is currently working to improve Sonicsense’s ability to manage multiple objects in complex environments, integrating advanced tracing algorithms. In perspective, an evolution of the robotic hand design is expected, with the aim of developing more sophisticated robotic hands, capable of performing tasks that require a more refined sense of touch. In addition, the team explores the possibility of integrating other sensory methods, such as the detection of pressure and temperature, to further enrich the interactions.

The work on Sonicsense represents a significant step towards the expansion of the sensory abilities of robots, bringing technology to a new level of complexity in interactions with the environment.