Digital Theses Archive


Tesi etd-02072017-090552

Type of thesis
E-mail address
Scientific disciplinary sector
INGEGNERIA - Biorobotics
relatore Dott. LASCHI, CECILIA
  • adaptive robots
  • ehd pumping
  • embedded actuators
  • evolutionary robotics
  • fluidic actuation
  • mathematical modelling
  • neural networks
  • nonlinear mechanics
  • soft actuation
  • soft robotics
  • stretchable mechatronics
  • underwater robots
  • untethered robots
Exam session start date
This thesis advances the field of soft robotic actuators, focusing in particular on soft fluidic actuators. The goal of my PhD has been to contribute to improve the adaptivity of soft robots through the use of fluidic actuation. In the first part of this thesis, I study the interfaces between the mechanics of the body, the control and the external environment. In the second one, I face one of the major challenges of soft fluidic actuation: the embedding of all of the components to obtain untethered devices. The technologies were investigated from all points of view: control, mathematical modeling, fabrication, and physical principles.<br>Following the methodology of bio-inspiration, I started by studying muscular-hydrostat systems in biological organisms; muscular hydrostats can be found in soft-bodied animals (e.g., cephalopods) as well as in organs with special functions of vertebrates (e.g., tongue in humans, trunk in elephants).<br>I developed a mathematical model synthesizing the mechanical features of muscular-hydrostats. I used the model together with genetic algorithms (optimization methods inspired by natural selection) in a novel computational experiment with the goal of understanding the key elements that allow these systems to achieve such an amazing dexterity while requiring very little computational effort.<br>The elements that emerged to be essential in the evolutionary shaping of the control strategy are: (a) the hydrodynamic added mass; (b) the elastic energy stored in the body; (c) the threshold in the power exerted by each muscle bundle; (d) the number of neural connections.<br>The knowledge of these elements not only contributes to a better understanding of the biology of cephalopods but can be used in the design of artificial controllers for soft robots. <br>In order to obtain part of the functionality of muscular-hydrostats with artificial soft actuators, I investigated Bending Fluidic Actuators (BFAs), composite structures made of hyper-elastic materials and reinforcing fibers, actuated by the pressure of a fluid in an internal chamber. These actuators recently received high attention in the robotics community due to robustness, simple fabrication and compliance. They can bend over 270 degrees and interact safely with the external environment.<br>I developed a novel mathematical model for BFAs, which for the first time is able to capture their nonlinear response and its variation with the design parameters.<br>The model describes the 3D large deformations of these structures by means of nonlinear continuum mechanics and a hyper-elastic material model. The computational complexity is kept low, requiring ~ 0.05 s of computation on a standard laptop PC.<br>This model is experimentally validated and can be used for both designing and controlling BFAs.<br>With the goal of proving the applicability of BFAs in mobile robots, we designed the FASTT robot, a hybrid-soft legged robot able to adapt to different environments by actively changing its morphology. The FASTT robot relies on self-stabilized dynamical cycles provided by its mechanical design and the dynamical properties of BFAs, not requiring any complex control algorithm.<br>The robot is simple and low-cost (&lt; 200 EUR); it achieved a speed of 119 m/h on flat ground (correspondent to a relative speed of 1919 bl/h) and was able to negotiate uneven terrains such as grass, spiky and soft grounds.<br>The second part of this thesis contributes to the embedding of fluidic technologies into soft robotics and wearable devices. Up to now, traditional pumps and compressors have been used to power soft fluidic actuators. These components are bulky and complicated, thus reducing most of the advantages provided by soft technologies.<br>Together with Shingo Maeda (Smart Materials Lab, Shibaura Institute of Technology, Tokyo), who was visiting Professor at the BioRobotics Institute, I investigated an innovative physical mechanism to transmit the energy from an electric field to a fluid through body forces, called ElectroHydroDynamics (EHD). The main advantage of EHD over traditional pumping technologies is the possibility to convert the energy without any moving mechanical part. Without the need of motors, bearings and lubrication, EHD allows an extreme simplification of the pumping systems and the fabrication using stretchable materials.<br>For a first proof of concept of the EHD principle in robotics, I developed a novel untethered self-sailing robot at millimeter-scale, powered by an external electric field. This robot is completely passive and it harvests the energy from its environment.<br>The weight of the robot is less than 2 grams. We registered a reliable speed of almost 4 cm/s, corresponding to ~1.2 bl/s, while absorbing very small currents ( ~10 uA) and powers (~ 10 mW).<br>Finally, I used the EHD principle to develop the first stretchable pump without any moving part. This project was developed during a visiting period at école polytechnique fédérale de Lausanne (EPFL), hosted by Prof. Dario Floreano (Laboratory of Intelligent Systems) and Prof. Herbert Shea (Microsystems For Space Technologies Laboratory), where we could find a convergence between the physical principle of EHD and the fabrication technologies for stretchable mechatronics.<br>I tested the prototypes and obtained for each unit a pressure of 340 Pa at 10 kV of applied voltage, absorbing a current of ~ 10 uA and a power of ~10 mW; the correspondent pressure densities are already comparable with commercial turbo-pumps: ~ 40 kPa/kg, ~ 60 Pa/cm^3.<br>The developed pump is the first of its kind and proved to be robust, scalable and silent. It can find applications in soft robotics, soft wearable technologies and stretchable mechatronics in general, transferring all the advantages of fluidic actuation to these fields.