Modeling Using Template Collections Current representation: Current search method: discrete search for closest template continuous search for best parameters Expansions to Robot design template parameters that allow more complex geometry discrete parameters constraints that include 2D design improve representation of articulation

Combining Templates User interface Inferring Connectivity Goals: select components alter template parameters set position and orientation Inferring Connectivity Goals: user-specified articulation structural stability Methods inference physical simulation

Scripted folded quadrotor design Goal: Automatically generate robot design given functional description Quadrotor(payload=10g, max_speed=1m/s, minimize: weight); Outputs DXF file for laser cutter, STL file for 3d printer, bill of materials for actuators / controls Current status: Python scripts generate structural elements by specifying geometry Quadrotor(diameter=20cm, thickness=1cm, motor_diameter=1cm, …); Parameterized dimensions, hardcoded topology Research Plan: Measure mechanical parameters of folded structural elements Generate simple models relating critical parameters to geometry Use models to determine required robot dimensions Aside: partition overall design space between folding / 3d printing / etc.

Rigid Folded Structures John Romanishin Rigid Folded Structures Attempt to develop a process to quickly create functional components by using a composite laminate of different 2-d laser cut layers. Potential Benefits: Quick to assemble Electronic components have support Clear distinction between flexible joints and rigid members Current design attempts are trying to develop a rigid quad rotor frame. Rigid layer is acrylic Flexible layer is adhesive backed thin peek. Functional details are added by simple FDM 3d printed components

Fabrication – Printable PCBs Copper Laden Ink: CO2 Laser (10um): 0.100in dia holes Nd:YAG Laser (355nm): 0.050in dia holes Thermal Transfer Printer: Mechanical Punch: 0.040 & 0.125in dia holes

REACT Programming Paradigm & Language Design Programming Paradigm & Language Designing and developing interactive systems Simplified programming model Facilitate event-driven and concurrent programming Implementation Source-to-source compilation framework (OCaml implementation) Target languages: Lego Mindstorms NQC Foldable robot C-subset Future Plans: Code generation framework that provides: Code free of common concurrency issues Automatically propagates data updates to relevant clients, ensuring data consistency Tool suite: Testing and static analysis techniques for ensuring functional correctness

Printable Robots – Fabrication (Harvard/MIT) Design Composition (gripper/crawler) Crawler design evolution Design Templates Stair climbing Printed robots playing chess Self-folding inchworm robot

Current Results Next goal Develop platform and algorithms for climbing Autonomous printed Hexapod 2-DOF legs Variety of gaits Tripod Wave Ripple Rotating Tripod with curvature Next goal Develop platform and algorithms for climbing

Mechanical Properties of Printable Processes Goal: To identify the mechanical properties of different materials and geometries. Mechanical Properties to identify Stiffness: Young’s Modulus Shear Modulus Strength: Yield Strength Ultimate Tensile Strength Materials 5 mil, 7 mil, & 10 mil polyester PEEK 3D printed materials Composites: Acrylic + polyester Geometry Triangular & Rectangular Beams Different Perforations Procedure Instron Tension Test: Measure Young’s Modulus, yield strength, and ultimate tensile in tension Instron Torsion Test: Measure shear modulus, yield strength, and ultimate tensile in torsion Image available at: https://alliance.seas.upenn.edu/~medesign/wiki/index.php/Courses/MEAM247-11C-P2P1