Abstracts Track 2020

Area 1 - Intelligent Systems

Nr:	7
Title:	Intrinsic Intelligence of Stimuli Responsive Polymers for Photonic and Robotic Devices
Authors:	Sara Nocentini, Isabella De Bellis, Daniele Martella and Camilla Parmeggiani
Abstract:	Stimuli responsiveness and reversible deformation are the peculiar characteristics of liquid crystalline elastomers (LCEs) that make them as highly suitable candidates for smart devices. The liquid crystalline properties and the elasticity of the polymeric network result in responsive soft materials able to sense the environment and act in response to selected stimuli for intelligent sensors, and actuators in photonic, robotic and medical applications. The intrinsic intelligence of these materials has been investigated and exploited to fabricate millimetric optical sensors as well as a microrobotic hand. On one hand, a temperature responsive LCE actuator has been combined with a Morpho Menelaus butterfly wing in a biotic-abiotic system. In response to environmental temperature changes, the nature optimized blue structural color of the wing is tuned by the reversible deformation of the integrated LCE actuator. Two different strategies have been proposed to control the visual sensor: a macroscopic deformation of the combined system induces an iridescence variation, whereas a nanoscale contraction generates a color shift through the lamellae interspacing variation, parameter that determines the structural coloration De Bellis et al., 2020. On the other hand, using the lithographic technique of direct laser writing, LCEs together with glassy polymers have been patterned into a micrometric design that resembles a humanoid hand. Controlling the liquid crystal elastomer alignments in the microstructure is possible to obtain four micro fingers that bend towards one single point thus mimicking the functionality of the human hand. Such microrobotic structure is therefore able to close and grab micro objects under a light stimulus. More interestingly, thanks to the opto-thermal responsiveness of LCEs, the microhand can operate even autonomously and catch colored micro-objects depending on the target optical properties Martella et l., 2017. These implementations show as a material intrinsic intelligence can be programmed and obtained in LCE based devices able to take simple autonomous decisions. De Bellis, I., Ni, B., Martella, D., Parmeggiani, C., Keller, P., Wiersma, D.S., Li, M.H. and Nocentini, S., "Color Modulation in Morpho Butterfly Wings Using Liquid Crystalline Elastomers". Adv. Intell. Sys. 2020, 2000035. Martella, D., Nocentini, S., Nuzhdin, D., Parmeggiani, C. and Wiersma, D.S., "Photonic microhand with autonomous action". Adv. Mater. 2017, 29(42), 1704047.

Nr:	8
Title:	Goal-driven Explainable Reinforcement Learning in a Robotic Scenario
Authors:	Francisco Cruz, Richard Dazeley and Peter Vamplew
Abstract:	In this work, we propose 3 explainable reinforcement learning methods to compute the probability of success as an alternative for a robot to provide human-like explanations to non-expert end-users. The methods obtained similar results in a robot navigation task but using a considerable different amount of resources.

Nr:	9
Title:	A Review of the Deep Learning Models for the Automatic Segmentation of Organs in Computed Tomography Images
Authors:	Malvika Ashok and Abhishek Gupta
Abstract:	Medical imaging is done with different modalities like MRI, CT, PET, ultrasound etc. Segmentation of organs in the medical images help the doctors in planning the treatment in lesser time and with higher efficiency. Results of manual segmentation vary from experts to experts and it is very time taking task. Automatic segmentation is the solution to the problem as it gives precise results. Although segmenting the organs automatically is quite challenging. Various techniques till now have been addressed in the literature for the automatic segmentation of organs in the medical images. Out of those deep learning models outperformed in the automatic segmentation of organs by giving precise accuracy. In this paper various deep learning models have been discussed briefly and two deep learning models were implemented on the computed tomography images and accuracy of the two models was calculated. At the end the results of the two models were compared to select the best model for organs segmentation in computed tomography images.

Area 2 - Robotics

Nr:	2
Title:	Communication and Interaction between Humanoid Robots and Humans
Authors:	Arbnor Pajaziti and Xhevahir Bajrami
Abstract:	This study is devoted to answering how will future robots be developed to assist or replace humans: generalists for everyday tasks, as humanoid robots suggest, or as barely perceptible, distributed elements of an intelligent environment? Presumably, there will be a middle ground between different types of humanoid robots, which build on the strength of their field of application. Nowadays, intelligent robots are possible, but in the long term, they will not surpass the people in their creativity, their ability to learn in their differentiation, and maybe not even catch up with them. So people still have a firm grip on the main switch.

Nr:	5
Title:	Photodeformable Artificial Muscles for Robots in the Micrometric Scale
Authors:	Daniele Martella, Camilla Parmeggiani, Sara Nocentini and Diederik Wiersma
Abstract:	Manipulating objects at the micro- and nano-scale is an open fascinating challenge that scientists are addressing by proposing different approaches, obtaining machines with basic or complex functions. Combining shape-changing polymers that respond to optical stimuli with 3D structuration at the microscale, we demonstrated synthetic microrobots entirely powered by light with a non-invasive and remote control. The arbitrary design allows to reproduce diverse animal and even humanoid tasks as walking and swimming but also the ability to grab and manipulate objects - overcoming natural limitations present at such small scale. The devices have been realized by Liquid Crystalline Networks (LCNs) which allow to perform different movements depending on their molecular alignment and, controlling their elastic deformation by light, wireless activation of the micro-machines is obtained. 3D patterning of LCNs is enabled by the use of Direct Laser Writing, a lithographic setup which allows to integrate different polymeric materials in the same devices with a resolution in the nanometric range. A micro walker was demonstrated to advance on different substrates; a micro swimmer to be prompted in liquids by structured light; and a micro hand to catch micro objects by external light control and even autonomously, depending on the target optical properties. The force developed by our photoresponsive polymers has been measured in response to different light power demonstrating how these materials can effectively work as artificial muscles, reproducing the force generation of biological muscles. All the aspect related to the material characterization, starting from the fabrication procedure to the mechanical behavior of polymeric materials will be presented to demonstrate how, starting from simple mesogenic monomers, it is possible to create polymeric microrobots with different abilities.