Tactile sensors for robot bodies

Tactile sensing technology has been studied for a long time, and a significant number of technologies have been proposed in the literature. This technology has also found growing relevance in the robotics field, including robot manipulation and safe human–robot interactions, where force/torque sensors are frequently used. Despite their accuracy, these kind of sensors do not allow dealing with complex physical interactions as in the case of multiple contacts (where only the resultant force would be detected, and therefore information on internal forces and contact locations are lost) or when major internal forces arise. Conversely, skin-like sensors used to measure and process distributed contacts, allow the reconstruction of the pressure distribution applied to the entire contact area. The goal of this thesis is to investigate the design of skin-like capacitive sensors for robot bodies, based on inkjet printing technique. Tactile sensors are devices based on the transduction of a component due to induce contact stresses, and they are often built as matrices of sensitive interconnected elements called taxels. The challenges for the manufacturing of tactile sensors are related to the maximization of their dynamic range and sensitivity, and to the use of a fabrication method that allows to optimize the sensor placement especially on curved surfaces, with the embedded electronics. To enhance the sensors dynamic range and sensitivity, the approach of exploiting the sensors vertical direction has been adopted, by creating stacks of capacitors. The validation of the proposed mathematical model is conducted by means of finite element simulations and the effectiveness of stacked capacitors in sub-optimal configurations has been experimentally tested by using inkjet printing as the main fabrication technique. The use of printed electronics has been chosen since it simplifies the fabrication process of tactile sensors with respect to conventional fabrication methods and it contributes to overcome the difficulties arising in the development of tactile sensors for real robot applications. From the system-level in the thesis we have design a front-end electronics for data acquisition and early tactile data processing based on embedded microcontrollers which can do both the acquisition and transmission part and programmed with an event-based algorithm to test the reliability about the touch detection. Results show that the stacked capacitors exhibit an enhanced dynamic range and sensitivity with respect to common single capacitors, for a given sensors area budget, and this suggests a method for designing tactile sensors with higher spatial resolution, higher transduction sensitivity and dynamic range. Furthermore, it has been verified that the new architecture can be connected and integrated with the manufactured sensors, detecting touches, and showing the resulting output capacitances on the host PC with low noise. This is a low-cost solution which allows to increase the number of sensors from which gather data and process them, meaning that a large contact area of the robot can be covered by the sensors and its electronic components. The result is the design of an embedded system that combines the manufactured part with electronics, creating an optimal blueprint for a future real implementation.