1. System-Level Design

2. Agenda Project Overview Customer Requirements Engineering Requirements
Use-Case Scenario
Functional Decomposition & Analysis
Concept Generation - Morphological Table
Feasibility Analysis
Benchmarking
Concept Selection
Risk Assessment
Current State
Action Items

3. Project Overview Fifth TigerBot. Human-sized: >5 ft
Ideally be used as a tour guide for RIT and as an inspiration to the younger generations.
Objectives: working prototype by 2015’s ImagineRIT.
Meet specs: able to walk, untethered, avoiding obstacles, able to get up from falls, responding commands.

4. Customer Requirements

5. Engineering Requirements

6. Use-Case Scenario

7. Functional Decomposition

8. Functional Decomposition

9. Functional Decomposition

10. Functional Decomposition

11. Morphological Table
Cyclical Drive

12. Morphological Table
Plumb Bob

13. Preliminary Feasibility Question
Feasibility Analysis
Preliminary Feasibility Question
Priority (U, H, M, L, P)
Assumptions
Can components/limbs handle weight?
Urgent
Square tube construction
Can we support untethered operation for 2+ hours?
Low
TBD
Can servos/gearboxes handle required torque?
Flexible budget, power is non-issue, size limitation based on joints
Can we get custom components made in timely manner?
Medium
Component design finalized
Separate or unified power supplies?
High
Rough idea of power requirements
Single stage or multistage voltage regulators?
Power requirements known
Can we prevent overheating of electrical components?
Urgent-High
Power requirements known, circuit design finalized
Can we prevent overheating of servos?
Power requirements known, torque requirements known
Can we support human voice interaction?
Preference
Sufficient power, CPU operational w/ ROS
Can we effectively balance?
Frame fully constructed, sufficient CPU speed, sensors interfaced
Can we avoid obstacles?
Able to walk/balance, CPU w/ ROS, sensors interfaced
Can we walk?
Able to balance, CPU w/ ROS, sensors interfaced
Can the robot detect its location?
CPU w/ ROS, external support network, sensors interfaced
Can the the robot detect its orientation?
CPU w/ ROS, sensors interfaced
Can the robot provide real-time data feedback?
Medium-Low
CPU w/ ROS, external support (including MSD team)

14. Torque Analysis

15. Benchmarking

16. Benchmark - TigerBots

17. Benchmark - Batteries

18. Benchmark - Servos

19. Teknic Demonstration Clearpath Integrated Servos
Great Control on the torque
Changes in fractions of a second
Click Me

20. Benchmark - Materials

21. Concept Selection

22. Unified vs. Separate Power Supplies
Unified: One battery pack for all power requirements
Separate: Different packs, one to handle high current draw for servos, and one to handle lower draw for computing and sensors.

23. Unified vs. Separate Power Supplies
LM317 regulator chosen
Very common, easily replaced
Can take max input voltage of 40VDC
Servos will be run ~48+ VDC
Unified PSU is not feasible

24. Single vs. Multi-stage Regulators
Single: Linear regulator driving CPU, sensors
Multistage: Linear regulator driving current source for CPU, sensors
LM317 drives 1.5A
BeagleBone draws up to 2A
OPA549
Power op-amp
Drives 8A (five times LM317)
LM317 to set threshold for OPA549, OPA drives current
Multistage more advantageous

25. High Torque Areas (Hips and Legs)
Teknic ClearPath-MCPV (Precision Position Control)
Absolute Positioning
Pulse Burst Positioning
Paired with select gearboxes will allow required torque
Absolute Positioning allows 4 preset positions
Pulse Burst Positioning allows precision positioning with speed and acceleration control

26. Low Torque Areas (Arms)
McMaster-Carr DC Gearmotors
Don’t require a gearbox
Can be set parallel to the joints or perpendicular to the joints.
For use on shoulder and elbow joints.

27. Materials Either Carbon Fiber Composites or Kevlar Epoxy
Carbon Fiber is light but prone to break from shock.
Kevlar resists shocks

28. Small Limbs/Joints Utilize Servos for each required DoF.
Upper body has smaller load
Use for arms and head.

29. Large Limbs/Joints
Toss up between worm gears, cycloidal gears, high torque servos and planetary gears
Points of interest where they are under heavy load
Use for hips, thighs , knees and ankles.

30. Hip / Torso Cycloidal Gear Zero Backlash HighTorque
Great for balancing the hip
Reduces energy use.

31. Foot / Ankle Plates for Foot Not Human Like Energetically Inefficient
Anthropomorphical Foot
Adopted Human oriented design
Highly efficient
Permits to increase foot clearance during leg swing.

32. Thigh / Leg Shock Absorbers Placed strategically at the knee
Relieves stress at the hip
Sustains the weight of the robot
“Human-like”-works like muscles

33. Connectivity WiFi or Bluetooth Wifi has longer range
Bluetooth has lower power consumption

34. Voice Interaction
Use pre-existing software and code from previous tigerbots.
Seems to work really well
More time/resource to focus on locomotion

35. Risk Assessment Risk Cause Effect Severity Detect Likelihood Total
Mitigate
Ownership
Budget
Overspending
No money 3 1 9
Ask Sahin for money
MSD Team
Manpower
Not enough engineer
Incomplete Robot
Increase Man Hours
Collapse
Design Failure
Injuries
243
Harness/Padding
RIT
Does not walk
Failed CR 27
Pass off to next MSD
Battery Shorts
Wiring
Fire 81
Triple check/ Insulation EE
Immobilized
No Power
Difficult to Move
Back up power supply (manual override)
Theft
Loose Security
No Robot
Locking LAB
Loss of Communication
Communication Failure
Loss of Control
Wired back-up
CE/EE
Components Breaking
Too much stress
Component Failure
Ansys Analysis ME
Logistics/Delays
Unknown delays
Gets pushed back
Put in extra hours to catch up

36. Current State Feasibility Limb Design Joint Design Skeleton Design
Power Calculations

37. Action Items Finalize all designs 3D model of skeleton
Simulation in Webot with skeleton
Spec out servos,gearboxes and batteries
Ansys Structural Analysis

38. References
Luo, Ren C., Chwan H. Chen, Yi H. Pu, and Jia R. Chang. "Towards Active Actuated Natural Walking Humanoid Robot Legs." IEEE Xplore. IEEE, June-July Web. 02 Oct
Takanishi, Atsuo and Kazuko Itoh. “Some Issues in Humanoid Robot Design”. IEEE Xplore. IEEE, Web. 02 Oct. 2014
Gini, Giuseppina, Umberto Scarfogliero, and Michele Folgheraiter. "Human-Oriented Biped Robot Design: Insights into the Development of a Truly Anthropomorphic Leg." IEEE Xplore. IEEE, 10 Apr Web. 02 Oct

39. Questions? Why is the robot angry??
Because someone kept pushing his buttons.