Accurate and fast contact detection between a robot manipulator and objects is crucial for safe robot-object and human-robot interactions. Traditional collision detection techniques relied on force-torque sensors and Columb friction cone estimation. However, the strain gauges used in the conventional force sensors require low-noise and high-precision electronics to deliver the signal to the final user. The Signal-to-Noise Ratio (SNR) in these devices is still an issue in light contact detection. On the other hand, the Eccentric Rotating Mass (ERM) motors are very sensitive to subtle touch as their vibrating resonant state loses immediately. The vibration, in this case, plays a core role in triggering the tactile event. This project's primary goal is to use generated and received vibrations to establish the scope of object properties that can be obtained through low-frequency generation on one end and Fourier analysis of the accelerometer data on the other end. The main idea behind the system is the phenomenon of change in vibration propagation patterns depending on the grip properties. Moreover, the project's original aim is to gather enough information on vibration feedback on objects of various properties and compare them. These data sets are further analyzed in terms of frequency and applied grip force correlations in order to prepare the ground for pattern extraction and recognition based on the physical properties of an object.
Keywords: active vibration sensing; contact detection; tactile sensing.