Davide Spinello

Dynamical systems and control: modeling and applications to robotics

Modeling and control of flexible systems: exploration and sensing

  • Bio-inspired locomotion and shape morphing: adapt to different terrains and sense material properties

Left: Giant millipede (longer than 20 cm) from south bank of river Periyar near Kodanadu. (By Irvin calicut (Own work) [CC-BY-SA-3.0 (http://creativecommons.org/licenses/by-sa/3.0)], via Wikimedia Commons.) Right: modeling as beam on elastic foundation.

  • Autonmous non-destructive monitoring of gas pipelines: motion control to collect data

The robot Explorer II (source: http://www.rec.ri.cmu.edu/projects/explorer/).

Simulation of a the closed loop dynamics across a pipeline elbow.

  • Autonmous non-destructive monitoring of gas pipelines: extract information from collected data to detect defects

Raw sensory data (Figure published in Sheiki F., Spinello D. and Gueaieb W., 2015).

Data mapped into an information entropy space to detect critical regions (Figure published in Sheiki F., Spinello D. and Gueaieb W., 2015).

Networked multi-agent systems: coordination through cooperation

  • By exploiting cooperation in multi-agent systems one can take advantage of synergistic effects that result in emerging collective behaviors
  • Information sharing among agents is crucial for actions coordination
  • Possible operational limitations (i.e. low bandwidth acoustic communication channel availability, geographic/environmental/tactical constraints) require specific algorithmic design

Simulated multi-agent system configuration in an area coverage mission with diffusive risk density. Agents track centroids of generalized Voronoi cells.

Simulated multi-agent system configuration in an area coverage mission with risk density locally perturbed by moving threats in the perimeter. Agents track centroids of generalized Voronoi cells.

Simulated cooperative target (black dot) state estimation by three mobile sensors with occasional communication. Trajectories generated by the motion control algorithm emerge from the minimization of the uncertainty associated to the estimate of the target state.

Scalar measure of the uncertainties associated to the target state estimates. Information sharing and structure of the underline communication network determine the convergence to a common value.

Dynamics of coupled multiphysics systems

  • Multiphysics Systems abound in science and engineering applications: examples are classical fluid solid interactions, micro- and nanoelectromecanical sensors and actuators, and multifunctional micro-composite materials
  • The thorough understanding, prediction, and control of multi-physics systems behavior require knowledge of systems dynamics and interaction among different physical worlds, e.g. thermal, hydro, mechanical and electrical

Micro- and nanoelectromechanical systems (MEMS and NEMS)

  • Mechanics: deformable body suspended above a rigid body + interaction through Coulomb force and short range forces (van der Waals - Casimir): nonlinear spring
  • Electromagnetics: variable gap capacitor, perfect conductors
  • Pull-in instability: there exist upper bounds of the external loads for which effects of nonlinear interactions (Coulomb, Casimir, van der Waals) overwhelm elastic restoring effect in the deformable body, and the deformable body eventually snaps and touches the lower rigid plate

Lumped model to illustrate the pull-in phenomenon.

Bifurcation diagram of the lumped mass-spring system

Continuum thermomechanical systems with meshless numerical methods

  • Meshless numerical methods (meshless local Petrov-Galerkin, smoothed particle hydrodynamics, local boundary integral equation...): as opposed to classical techniques such as finite elements and boundary elements, the only structure required is a set of scattered nodes
  • In some situation meshless methods can alleviate problems associated to classical methods (e.g. meshing and re-meshing, element locking and distortion in finite elements)

Wave propagation and heat conduction in multimaterial solids: modeling of interfaces (jump of the gradient) in meshless numerical methods

Traveling stress wave associated with a sinusoidal load of finite duration applied at the free end (left) of the cantilever. The interface is located in correspondence of the red dot

Displacement of a bimaterial cantilever associated with a sinusoidal load of finite duration applied at the free end (left). The line at the bottom represents the initial length with interface located in correspondence of the red dot

Shear bands in elastothermoviscoplastic materials: nonlinear transient problem with localized instability phenomenon

Space distribution of the shear stress in a block of elastothermoviscoplastic material in a simple shear test: an elastic unloading wave emanates from the specimen center after the shear band formation, revealed by the sudden drop of the shear at the center

Time history of the shear stress at the center of a elastothermoviscoplastic material specimen in a simple shear test: the sudden drop corresponds to the shear band initiation and formation