Being able to cover the entire body of a robot with soft tactile sensors has become an attractive concept in intelligent robotics. Soft, stretchable materials can conform around surfaces and also absorb impacts, which is beneficial for safety around humans and in unpredictable environments.
However, such a sensor has not yet been developed because such systems are challenging to design and manufacture. In the traditional approach to tactile sensors, a high number of transducers must be placed on soft and stretchable substrates and then wired together. This strategy results in large wiring bundles, rigid elements, and many potential points of failure.
A geophysical imaging method called Electrical Resistance Tomography (ERT) has recently shown promising results in tactile sensing, simultaneously achieving large-area coverage, ease of manufacturability, and robustness. ERT is a reconstruction method that estimates the internal conductivity distribution of a conductive medium using electrodes that are typically located only around the boundary. These electrodes are used to inject electrical currents into the medium and also measure the resulting voltage potentials. The internal conductivity distribution is reconstructed from these measurements and calibrated to external stimuli.
This project aims to develop a fabric-based ERT tactile sensor that can cover a large area. To achieve this goal, we focus on two aspects. First, we develop a low-cost and reliable sensor manufacturing method that involves fabric layers and sewing. Second, we focus on overcoming the limitations of conventional ERT-based sensors to improve sensing resolution and framerate.
In the future, we plan to optimize the sensor design and conductivity reconstruction techniques to improve tactile sensor's capabilities. Overall this project aims to provide a systematic method to develop whole-body tactile sensors for any robot.