Analysis of the Foot Placement Estimator

Transcript

1. Foot Placement Estimator A Dynamic Measure of Balance for Bipedal Robots ECE 688: Course Project Safwan Choudhury
2. Biped Locomotion
3. Motivations ★ Navigating Human Environments ★ Space Exploration / Military Applications ★ Rehabilitation / Prosthetics
4. 2011 Honda ASIMO at FIRST Robotics ★ Navigating Human Environments
5. 2011 Honda ASIMO at FIRST Robotics ★ Navigating Human Environments
6. NASA’s Robonaut for International Space Station (ISS) ★ Space Exploration / Military Applications
7. Exoskeleton by Ekso Bionics ★ Rehabilitation / Prosthetics
8. Exoskeleton by Ekso Bionics ★ Rehabilitation / Prosthetics
9. Achieve stable and energetically efficient walking control strategies for bipedal robots which mimic human-like gait
10. Walking Control Strategies ★ Zero-Moment Point ★ Passive Dynamics ★ Biologically Inspired Techniques ★ Central Pattern Generators ★ Neural Networks ★ Zero-Hybrid Dynamics
11. Zero-Moment Point
12. Zero-Moment Point ★ Introduced by Vukobratović in 1968 ★ Point where the inertial and gravity moments cancel out (i.e. the moment is zero). ★ Stable if this point is within the support polygon.
13. Support Polygon (Region of Foot Support)
14. 3D Linear Inverted Pendulum ★ Models entire biped as a single point mass located at the center of mass (COM)
15. Remarks on ZMP ★ Popular but energetically inefficient ★ Statically stable walking only ★ Very “robotic” (not human-like gait)
16. Passive Dynamics
17. Passive Dynamics ★ Introduced by Tad McGeer in 1990 ★ Powered only by: ★ Force of Gravity ★ Natural Dynamics θ1 θ2 m m mH x z φ
18. Tad McGeer’s Seminal Work (1990) ★ Passive Dynamic Walker
19. Tad McGeer’s Seminal Work (1990) ★ Passive Dynamic Walker
20. Limit Cycles and Stability in Passive Gait (Goswami et al) ★ Passive Dynamics Limit Cycle
21. Virtual Passive Dynamics ★ Emulates the force of gravity as a small actuator torque θ1 θ2 m m mH x z φ
22. Remarks on Passive Dynamics ★ Highly energetically efficient ★ Not robust (even small perturbations) ★ Very specific initial conditions
23. Foot Placement Estimator
24. Foot Placement Estimator ★ Introduced by Derek Wight in 2008 ★ Aims to restore balance, instead of constantly maintaining balance.
25. Brief Summary of Derivation
26. Impact of FPE Location
27. FPE State Machine (Dwight, 2008)
28. Yet Another Biped Robot (YABR) ★ 2D FPE-Based Gait
29. Yet Another Biped Robot (YABR) ★ 2D FPE-Based Gait
30. Remarks on FPE ★ Inherently robust (by design) ★ Can combine with simple control strategies ★ Makes a few assumptions (i.e. mass-less legs) ★ Requires fast control + actuators
31. Contributions to Bipedal Locomotion ★ Can be combined with any other approach as an emergency mechanism (even ZMP). ★ Provides an indicator of how unstable the biped is. ★ Biologically verified with human experiments
32. MASc Research + Project
33. Project Goals ★ Implement FPE-based control strategy in dynamic simulations ★ Constrain sagittal motion without the boom (!)
34. Dynamic Simulations of 3D FPE Controller
35. Dynamic Simulations of 3D FPE Controller
36. Dynamic Simulations of 3D FPE Controller
37. Dynamic Simulations of 3D FPE Controller
38. MASc Goals ★ Virtual Passive Dynamics ★ Foot Placement Estimator ★ Minimizing Energy Consumption ★ Human-Like Gait Cycles
39. Questions?