Embedded Files
Harvard Microrobotics Lab Research (2018-2025)
Harvard Microrobotics Lab Research (2018-2025)
Brief overviews of lab research projects and contributions below - see publications to read full published papers and dissertation for more details.
Scaling Microrobots - Streamlining Fabrication
Challenge: Fabricating insect-scale robots is a delicate process which gets becomes even more challenging at even smaller scales. During initial runs to make a half-scaled robot, we had <10% yield rate of actuators (~1cm long, 8 required per robot, multi-day process), hindering ability to reasonably fabricate a smaller scale robot.
Action: improve actuator fabrication process
identify failure modes of existing actuators - including mechanical delamination and breaking at multiple stages of fabrication and electrical disconnection/shorting of the ground/signal/bias channels
alter CAD layer designs and tune laser cut parameters and recipe to reduce the likelihood of breakage and shorting
increase cleaning processes and implement material quality checks in process for freshness of carbon fiber to reduce risk of layer delamination
Result: >90% actuator yield, enabling fabrication of half-scaled (2.2 cm, 320 mg ) version of the quadrupedal microrobot. Tested payload capability for this smaller design and enabled additional characterization.
A scaled (2cm) version of the original (4cm) laser-fabricated robot was developed.
Variety of failure modes for actuators at various stages of fabrication.
The laminate-based fabrication process improved and documented for future researchers.
Payload Improvement, Testing and Analysis
Challenge: Electroadhesive versions of the 1.5g ambulatory microrobot could be someday used for inspection of industrial components, but would need adequate mechanical payload capability to carry and climb inclines with onboard high-voltage electronics.
Action: Test existing capabilities and improved design options
tested shear force capabilities for 4 different types of electroadhesive footpads (with varying dielectric material thicknesses and fabrication method - both sputter coated and laminate layups)
collected shear data over a range from 300 to 700V to examine relative strength improvements at higher voltages
tested robot chassis capabilities via development of inclined angle 'pull-up' test for robots loaded with varying adjustable dummy payloads (0-2g added payloads, 3D printed)
Result: identified an option for improved electroadhesive performance (thinner dielectric, laminate layup fabrication), resulting in single pad performance capable of holding 50g (30x robot bodyweight) payload on these flat conductive surfaces. This work also contributes towards experimental understanding of relationship between greater voltage and improved shear forces, for consideration of payload tradeoffs (with greater voltage electronics packages requiring additional weight) for communication with collaborators.
Several of the electroadhesive pad recipe options tested
Shear force tests (left) and robot demo (right)
robot 'pull-up' tests with mock payloads (up to 2g)
Enhancing Electroadhesion for Climbing Microrobot
Challenge: Electroadhesion does not work as well on rough surfaces as flat surfaces. Since the strength of electroadhesion is closely related to the dielectric gap between positive and negative charges, rough surfaces increase that gap and reduce adhesion capability.
Action: Enhance the compliance of the electroadhesive pads to improve performance on rough surfaces.
designed a series of 'spoked' pad designs designed to bend and adhere closer to the surface, thereby improving adhesion
adapted script to generate a set of stochastically-defined 'rough' surfaces (defined by the distance between peak and trough, and spacing between features) on the same relative order of roughness as the pad thicknesses, 3D printed set and sputter coated in copper and silver coatings for electrical conductivity
implemented test setup and testing procedure with nano17 force torque sensor and motorized shear pull system to quantify and characterize individual pad shear force performance
fabricated set of electroadhesive pads and robot, ran video-captured robot-level climbing tests at increasing inclines (0-45 degrees) on surfaces of varying roughness with different pads
Result: compliant electroadhesive spoked pad designs for robust adhesion to rough surfaces and heterogeneous terrains, with a 48% increase in robot speed over a 75µm stochastically defined, 3D printed and sputter coated terrain.
Diagram of hypothesized effect of enhanced compliance, allowing closer surface conformation of electroadhesive pads.
Series of 'rough' surfaces defined, 3D printed, sputter coated, and tested.
Individual pad design and robot-level climbing tests conducted to evaluate performance.
Kirigami Patterns for Improved Compliance
Challenge: Enhancing electroadhesive pad compliance was shown to slightly improve adhesion performance in the work above, but could benefit from even stronger performance improvement and some degree of design understanding.
Action: perform a parameterized exploration of designs to better identify improved options for even higher adhesion on rough surfaces
defined a wide range of potential pad designs and performed a broad experimental sweep of options to identify better performing options
considered the relevant parameters required to design different patterns, and identified design limitations based upon fabrication constraints (border thickness limiting effective area of thinner designs)
developed a simple theoretical framework capturing the rough behaviors seen in the spoked designs, with adhesion pressures decreasing at the extremes of reduced compliance (thick) and reduced effective area (thin)
experimentally tested range of widths for both spoke and serpentine pad designs to compare against predictions
Result: designed an enhanced serpentine electroadhesive pattern that performed 3x better than a comparable baseline circular pattern on a rough (75µm ) terrain.
Electroadhesive pad design exploration and parameter consideration.
Experimental sweep and analysis
Theoretical framework predicting performance trends of pads at design extremes.
Real-World Demos
Much of the work completed here has been done considering the future potential of the robot to be used in real-world industrial inspection applications (aerospace/jet engines, automobile, wind turbines, etc.), but characterizations were primarily done with lab-generated surfaces (for consistency and repeatability).
We further demonstrated the capability for the robot to adhere to real-world surfaces, including a rough textured automobile caliper part, inclined plane machined parts, and high curvature / high complexity automobile transmission.
See videos section in publications to see them in action.
Robot walking on rough textured automobile caliper.
Electroadhesive pad adhering to automobile transmission component with high curvature.
Industry Project Examples (2014-2017)
Industry Project Examples (2014-2017)
Below are a few select examples of industry internship and coop work. Examples are limited to projects that could be described publicly, but see resume for brief details on wider range of details on industry experience.
Syringe Needle Study: [Testing/Analysis] | Regeneron (2015)
Syringe Needle Study: [Testing/Analysis] | Regeneron (2015)
I had the opportunity to work on a project characterizing forces and image data and angles for fluid from a needle.
Problem: The auto-injector devices used for dispense of drug product included a syringe component. The properties of uncapped syringe needles were being investigated for a drug product.
Action: Designed and ran test method to determine impact
tested over 100+ samples
experiment spanned 24 hrs at various timepoints
designed test fixture for Cognex camera Instron attachment in Solidworks and 3D printed with Connex
refined and developed detailed timing instructions for test method to ensure all samples tested at correct time
analyzed images with ImageJ to determine stream angles
ran minitab statistics on data to draw correlations/regressions, plot box plots and understand significance of results
presented results internally to full medical device group
Results: Drew conclusions on impact of uncapped syringes
test method developed to be utilized in future
potential impact on patient usage instructions
Images of uncapped syringe needles (pics taken in lab with toolmaker microscope)
Cognex camera / Instron Tensile Tester Needle test setup developed
Example "break-loose, glide-force" load profile collected from Instron
Scanning Electron Microscope (SEM) Filter Investigation [Testing/Analysis] | Regeneron (2015)
Problem: Filter pore size was assumed to match manufacturer specifications; a test of the actual pore size could help identify any potential causes of early filter clogging.
Action: Designed and ran test method using SEM
wrote SEM test method protocol
used 3 samples each from 4 different filter variations
sputter coated and prepared all filter piece samples
collected image data from SEM magnification
used imageJ to measure the pore sizes; averaged results for indication of average size pore size per filter
ran EDX (energy-dispersive x-ray) analysis with SEM to gather material / element composition data
Results: Drew conclusions on variability and pore size
no unexpected bioburden elements found through EDX
results to be used / compared against manufacturer specs
1000X magnification of the sample using SEM
2000x magnification of filter
Gasket Geometry Prototype: [Design] | iRobot (2015)
Problem: Manufacture of gaskets wastes a lot of material; a method of reducing the material needed to die-cut geometry out could save on costs.
Action: Designed prototype concepts for reducing material
researched existing /prior patents on space minimization
experimented with different geometries using simple paper prototypes to fold out from smaller footprint
identified most promising designs and modeled in PTC (Pro-E) Creo CAD program
created drawing PDF and converted to Corel file
laser cut prototypes out of various foams (adjusted the power and speed settings to identify optimal settings for the material)
shared findings with team engineers
Results: Functional prototype of redesigned gasket
80 % less material used with new design
to be considered for potential manufacture implementation
animation of gasket unfolding
design when initially cut out
design when folded out into actual usage form
Carrier for Large Robot: [Design] | iRobot (2015)
Problem: large, heavy, bulky robots can be hard to carry to field tests and events; having a dedicated carrier to make this process easier while protecting the robot would be beneficial.
Action: Designed and prototyped carrier device
measured and CAD-modelled shape of the robot itself
designed carrier through modification of an existing cart design to include cushioning and attachment points
prototyped the carrier
made modifications based on dimensions, fit and hole spaing requirements to support load
developed CAD PDF drawings to be sent to machine shop for metal prototype
Results: Prototype and CAD model/drawings
prototype serves as proof of concepts for design
next steps: machining of parts
side view CAD for padding design
carrier prototype
Test Bed for Store Display Testing [Design/Testing] | iRobot (2015)
Problem: testing required to ensure that store display units would showcase correct behaviors
Action: Built test bed and tested robot
planned test bed design based on marketing mockups of actual display designs
ordered wood and constructed test bed for robot, included easily configurable middle divider to test different sizes
tested and videotaped robot behavior during normal run
identified issues with certain configurations of test bed and materials used
communicated issues with product manager and worked with engineers to resolve
Results: built and tested display
identified potential issues to be resolved before sending
test beds displayed in almost 300 Best Buy stores nationwide during Roomba 980 launch
test bed created and used for testing.
actual public Best Buy store display.
Instron Force Testing [Design/Testing/Analysis] | Hologic (2016)
Problem: Disposable components were being tested to ensure they met requirements for pull force specifications.
Action: Ran test, modified fixture, suggested improvements
tested over 30+ samples
ran Instron method for tension and compression tests
analyzed data with minitab to determine if normal data set
modified test fixture model with Soliworks to reduce variation in Instron base plate attachment in order to reduce operator variation during testing
created drawing for machining of revised parts
designed grip modifications to increase ease of user pull in case higher forces unavoidable
Results: Ran and improved test method
results of testing will be used in specification reassement
redesign of grips to be considered as alternative
new fixture design to be machined and used in future tests
CAD for fixture plate redesign where the holes are dimensioned to fit the Instron plate for consistent mounting
Northeastern Capstone Design (2016-2017)
Northeastern Capstone Design (2016-2017)
Automated Robotic Cake Decorator
Goal: Design an automated solution for the home cook to automatically frost a cake based on an input design.
Action:
brainstormed various concepts for dispense mechanism
linear-screw piston-cylinder system selected as a optimal solution for frosting dispense: displacement rather than force mechanisms allows for consistent extrusion with varying viscosities
researched different options and performed calculations on motor torques and displacements to determine feasible component selections
testing of system, experimenting with different Arduino parameters for motor controls
tackle and solve problems that arose during process- such as minor deflection due to moments that caused slight bending in system
Results:
functional rosette extrusions
system that moves reliably without binding
goal - to serve a frosted cake for capstone day
Won 1 of 2 senior capstone awards for design and completeness of the project.
20170417_225127.mp4 Class Projects & Writing Examples (click paper corner to expand)
Class Projects & Writing Examples (click paper corner to expand)
Microfabrication laboratory [Class Lab Papers - Harvard 2019]
All work done in a collaborative lab setting, including hands-on cleanroom work at CNS (Center for Nanoscale Systems).
LAB1.pdfFabricating a Photodetector
LAB2_JSHUM.pdfWet Etch for Light Emitting Diodes
Lab3_MOSFETS_JSHUM.pdfMOSFET Fabrication
Lab4_JShum1.pdfMicrodisk Fabrication
MOSFET_Lithography_Review_Jennifer_Shum_old.pdfSoft Lithography Review
Materials Processing techniques and selection criteria [Class Papers - Northeastern 2017]
ProcessingThinSheets_591ed8_13645f85159547748d2f9fc5f44e8ab0.pdfProcessing techniques for flat sheets of different materials
LaserSintering_591ed8_64296c87d5e34f849759c3cb3922cc58.pdfDirect Metal Laser Sintering Review Paper Proposal
MatrixComposite.pdfSilicon-Cu matrix composite for use as conductive polymer seal (using CES selector tool)
Vitreloy.pdfUnique properties and processing of Vitreloy (an amorphous alloy of the BMG type)
Bioinspired Structural Color Explainer Article [Class Project papers - Northeastern 2015]
An article written to explain scientific topics to broader audiences - more specifically, on structural color and bioinspiration.
Butterflies.pdfCS/FEM [Class Project papers - Harvard 2018-2021]
A collection of final project papers in computer science/algorithms/optimization/simulation disciplines.
SHUM_STEINMEYER_cs289finalpaper (4).pdfAutonomous Cooperative Behavior with Limited Binary Sensing
Underactuated.pdfPlanning Paths and Poses for an Ambulatory Microrobot
CS286_Final_Report.pdfColony and Cockroach Swarm optimization for Travelling Salesman Problems
Torsion_DEA___ES128_Final_Report.pdfFEM Simulation of Torsional Dielectric Actuator Motion with Coil-Wrapped Core Geometry
Page updated
Google Sites
Report abuse