PowerPoint ******

```Control Design to Achieve Dynamic Walking
on a Bipedal Robot with Compliance
Young-Pil Jeon
Contents
1. Introduction
2. Robot Model
3. Control Algorithm
4. Experiments
5. Conclusions
Introduction
1. Paper Introduction
(Control Design to Achieve Dynamic Walking
on a Bipedal Robot with Compliance)
2. ZMP (Zero Movement Point)
Paper Introduction

main issue in the paper
 One of main issues in biped mobility is uneven terrain locomotion.

Approach


1. The ZMP based walking

2. dynamic(limit cycle) walking
Contribution
Main contrition is a development of an adaptive control for
generating dynamic locomotion under external disturbances
What is ZMP
•
ZMP is point that is become zero to sum moment (gravity, inertia)
ZMP
• ZMP based walkers have the following common
features
• 1. the stiff joints are position-controlled with high gain
• 2. ZMP criterion is used to guarantee biped walkers
not falling over
• 3. 3D target trajectories in the task space are first
planned under ZMP constraints and then solved the
“inverse problem” for joint command
Dynamic walking
• Dynamic walking has the following features.
• 1. the compliant joints are torque/force-controlled
• 2. stable waling is achieved by finding attractive stable
orbits called limit cycles.
Robot Model
1. A biped Robot with Compliant Actuators
2. Dynamic Model of biped robot
A Biped Robot with Compliant
Actuators
• The robot composed of 32 joints(except fingers with
40 joint)
• The robot itself is 1.5 meters tall and has a mass of
Fig2. A hardware prototype and a
kinematic model of robot
A Biped Robot with Compliant
Actuators
• Each leg has six joints
• Three pitch joints(hip, knee, and ankle pitch) are
actuated by cable-driven modules
• The other three joints(hip roll/yaw, ankle roll) are
actuated by harmonic drive modules
What is harmonic drive
• Harmonic drive is gear drive which is better size,
weight than general gear drive
• The one is composed of circular spline, wave generator
and flexspline
Fig3. harmonic drive
Fig4. composition of harmonic
A Biped Robot with Compliant
Actuators
• On the other hand, arm joints are actuated by gear
driven modules
• Lower body is actuated by joint torque controllers
using 12 torque sensors
• Upper body is actuated by position controllers
A Dynamic Model of the robot
• On the other hand, arm joints are actuated by gear
driven modules
• Lower body is actuated by joint torque controllers
using 12 torque sensors
• Upper body is actuated by position controllers
Control Algorithm
1. Gait Sequence Synthesis using Walking Motion States
2. Parameterized State Machine and Target Trajectory
Generation
4. Desired Control Torque Generation
Control Algorithm
Fig5.
Control Algorithm
6
Control Algorithm
7
Gait Sequence Synthesis using Walking Motion Sates
6
8
Parameterized State Machine and Target Trajectory Generation
9
Parameterized State Machine and Target Trajectory Generation
1) Adequate Foot Placement with a Capture Point
(1)
(2)
10
2) Step Motion Planning by Simple Interpolation
(3)
Desired Control Torque Generation
1) State Machine based Control
(4)
2) Swing Leg Gravity Compensation
Desired Control Torque Generation
3) Stance Leg Balancing with Virtual Model
(5)
Desired Control Torque Generation
4) Sensory Reflex Control with Potential Barrier
(6)
11
(7)
```