We investigate two main sensing modes: touch and vision, with the long term goal of progressively substituting most of the sensors of the iCub with their ED counterpart.In the visual domain, we work on the improvement of pixel functionality, noise resilience and size, and the development of data serialisation, a crucial step towards the integration of higher resolution sensors on the robot. We proposed a novel and more robust circuit for change detection in the visual signal, designed to tackle one of the major drawbacks of change detection, by filtering high frequency noise without low pass limiting the response to large and fast transients.The sparseness of tactile input over space and time calls for ED encoding, where the sensors are not continuously sampled, rather wake-up at stimulation. The iCub is currently equipped with capacitive sensors, at the same time, different groups within IIT are developing new materials and technologies for tactile transducers. This line of research aims at complementing such developments with neuromorphic ED readout circuits for tactile sensing, based on POSFET devices.
ED Touch Sensors – Ali Abou Khalil
We focus on the study, design and characterization of tactile sensors for low-latency compressive pressure sensing based on the use of a piezoelectric material coupled to mixed-mode neuromorphic CMOS circuits (Leaky integrate and Fire Neuron (LIF) and the Change Detector (CD)) based on event driven approach. The sensing element is based on the POSFET (Piezoelectric-Oxide-Semiconductor-Field-Effect-Transistor) device, whereby the polarization induced by pressure in the piezoelectric material biases a CMOS transistor. From one side, the sustained configuration based on the POSFET with the LIF recalls the sustained type of skin mechanoreceptors (Merkel cells) in humans where the firing rate of the LIFs is proportional to the pressure applied to the POSFET. On the other side, the transient configuration based on the POSFET with the CD recalls the transient activation of Meissner Corpuscles based mechanoreceptors where the output frequency of the taxel proportional to variations of the input pressure. Both sensor configurations encode the sensed force on the piezoelectric material into trains of digital pulses sent out off chip to the post processing unit.