Main ideas:

1. Main ideas: • Use novel layered prototypingmethods to create compliant biomimetic structures with embedded sensors and actuators (Cutkosky, Kenny, Full) • Develop biomimetic actuation and control schemes that exploit “preflexes” and reflexes for robust locomotion and manipulation (Kazerooni, Howe, Shadmehr, Cutkosky) 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

2. Shaft coupling Shaft Motor Leg links Building small robot legs with pre-fabricated components is difficult... Boadicea leg Electric motor/link 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

3. Concept design for a biomimetic “Insect-Leg” A prototype design of the same leg employing three-dimensional plastic “exoskeleton” surrounding with embedded actuators, sensor and cooling system. 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

4. Shape Deposition Manufacturing(SU/CMU) 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

5. SDM allows finished parts to be inserted at any point in the cycle Green link and red bearings are added as finished components 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

6. SDM capabilities • Slides and web pages of parts that would be difficult or impossible to create using conventional manufacturing methods • Topology that would be almost impossible with conventional machining tilted frame (CMU/Stanford) • Integrated assembly of polymers with embedded electronics and interconnects (CMU Frog Man) • other example parts from RPL at Stanford 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

7. MicroStructures and Sensors Lab (MSSL) Kenny Research on Fundamental Properties and Applications of MEMS-based MicroMechanical Devices. • Micromechanical Sensors. • Micromechanical Elements for Scientific and Technological Collaboration Partners. • Devices and Instruments for Studies of Fundamental Properties of Micromechanical Structures. Collaborators : IBM, JPL, NRL, SNL, SAIC, Medtronic, Raychem, Lucas, Seagate, Perkin-Elmer... Students from :ME, EE, Appl Phys, A/A Piezoresistive Lateral Accelerometer 2-Axis AFM Cantilevers for Surface Friction Experiments and Thermomechanical Data Storage Flow Visualization in Microchannels Ultrathin Cantilevers for attoNewton Force Detection

8. Embedded SMA actuators • Intial experiments with epoxy and urethane polymers and various sacrificial supportmaterials have underscored the need tobuild in disposable fixtures for proper alignment. Shape Memory Alloy wire with water cooling channels Epoxy acrylic 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

9. Approaches to design with layered shape manufacturing Usually people think of taking a finished CAD model and submitting it for decomposition and manufacture Example: the slider-crank mechanism, an “integrated assembly” built by SDM 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

10. SDM process planning: geometric decomposition for tool access build direction Cross section of part material (gray) in support material 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

11. Decomposition into ‘compacts” and layers • Several levels of decomposition are required Complete Part Compacts Layers Tool Path 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

12. Testing for compactness Z There exists no point, p, on S which is an inflection point with an undercut surface above an upward-facing surface. 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

13. Layers produced by automatic decomposer for slider crank mechanism Gray = steel, brown = copper support material 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

14. Layered shape deposition - potential manufacturing problems • finite thickness of support material • poor finish on un-machined surfaces • warping and internal stresses 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

15. Slider crank can be built entirely from two kinds of primitives Yellow = part material, blue = support material 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

16. Merge algorithm for compacts (Binnard) f (a,b ) 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

17. Truth tables for Boolean operations on compact lists P = part material S = support material c = f (a,b) 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

18. Building Designs from Primitives • Here is the result of building slider-crank from primitives • allows manufacturability analysis at design time 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

19. What gets sent to the Manufacturing Service What the Designer works with Primitives + Merging Rules SFF Object made up of Part and Support Compacts The Final Geometry Building a robot joint from a library of shapes 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

20. Design for a prototype pneumatic knee joint built from primitives (M. Binnard) Magnetic Gear Tooth Sensor Pneumatic Actuator Link 1 Link 2 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS

21. a) (top view) b) (side view) d d d(a1,a2) d(a1,a2) l 2l Dd Minimum gap/rib thickness Generalized 3D gap/rib e) (side view) 2l l d(m1,m2,m3) d(m1,m2,m3,a1,a2) Wc/l >= 2 m1 m2 m3 m1 m2 m3 Minimum feature thickness Comparison with VLSI approach SFF-MEMS VLSI Boxes, Circles, Polygons and Wires Decomposed Features SFF-MEMS Design Rules Mead-Conway Design Rules 1998 PI MEETING FOR LEGGED LOCOMOTION AND MUSCLE-LIKE ACTUATORS