ME Seminar on "Enabling Novel Robotic Capabilities with Granular Materials and Suspensions"


Monday, January 13, 2014 - 4:00pm to 5:00pm


ESB 1001


Anette (Peko) Hosoi, Massachusetts Institute of Technology, Dept. of Mechanical Engineering

ABSTRACT:  The intersection of soft matter and robotics abounds with opportunities.  In this talk I will discuss the coupling of robotics with (dense) particulate systems including suspensions, granular media, field-responsive fluids, and other particle-laden flows.  A distinctive characteristic shared among dense collections of particles – from suspensions to sand piles – is the fact that small changes in microstructure can lead to large changes in macroscopic properties. For example, removing a few particles from a granular arch may cause the entire structure to tumble, or reorganizing the microstructure in a dense suspension may increase the effective viscosity by many orders of magnitude. The key capability we require in order to harness these large nonlinear responses is the ability to control and rearrange particle microstructure in confined geometries. We are investigating three approaches to this goal using electric fields, magnetic fields, and mechanical forcing. The first two approaches rely on active fluids, namely fluids that change their microstructure in response to an applied field. 

In collaboration with Boston Dynamics, an industrial leader in the field of advanced robotics, we are investigating the incorporation of such fluids in hydraulic systems. In general, hydraulic strategies tend to be restricted to large-scale robots (e.g. Boston Dynamics’ famous “Big Dog” is the size of a small mule and weighs 240 lbs). This size limitation arises primarily from the the fact that mechanical servovalves are not easily scaled down due to their mechanical complexity and required fabrication tolerances. Owing to this complexity, small high-performance hydraulic servovalves can be extremely expensive (> $1000 per valve) rendering small-scale hydraulic robots economically infeasible for all but the most specialized applications. Active fluids may offer a simple solution to this technological barrier. By replacing the traditional hydraulic oil with an electrorheological fluid, complex mechanical valves can be replaced by two electrodes. Since electrodes are cheap, this advance brings the cost of valves down by three orders of magnitude potentially enabling a new class of small scale hydraulic systems. In this talk I will discuss our efforts to model, test, design, and prototype these small-scale valves, including dynamic visualization of particle chain formation, quantification of boundary effects (e.g. channel roughness), modeling bulk dynamics, and prototype development to demonstrate feasibility in real robotic systems.

HOST:  Prof. Eckart Meiburg

Event Type: