model predictive impedance control mpic. motor control features 1.feedback (closed loop)...

25
Model Predictive Impedance Control MPIC

Upload: frank-simon

Post on 20-Jan-2016

225 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Model Predictive Impedance Control

MPIC

Page 2: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint
Page 3: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Motor Control Features

1. Feedback (closed loop)2. Feedforward (open loop)3. Learning4. Predictive Control5. Joint (muscle) impedance6. Interaction with environment7. Hierarchical8. EPH, Rhythmic & Tracking movements, …

Page 4: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Limbic System

Associative Cortex

Cerebellum Motor Cortex Basal Ganglia

Spinal Cord

Musculo-Skeletal System

Musculo-Skeletal System

Movement

Motor Program

Need

Plan

Highest Level

Lowest Level

MiddleLevel

Page 5: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint
Page 6: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Trajectory

Selector

FeedforwardController

SpinalCircuits

andMuscles

Joint-LoadDynamics

Torque

Disturbance

ReceptorsDelay

Brain Model

Joint Movement

Td

d

Stiffness Control SchemeStiffness Control SchemeStiffness Control SchemeStiffness Control Scheme

Page 7: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Trajectory

Selector

FeedforwardController

SpinalCircuits

andMuscles

Joint-LoadDynamics

Torque

Disturbance

ReceptorsDelay

Brain Model

Joint Movement

Td

d

Stiffness Control SchemeStiffness Control SchemeStiffness Control SchemeStiffness Control Scheme

Page 8: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Feedforward Controller

Identifier

Brain Model

Delay

b

.

M P C and

Algorithm Adaptation

EMG

TorqueTorque

Joint-LoadJoint-Load

+ +

+ +--

Selector Trajectory

+

G1

G2

G3

+

System-Disturbance

Models

Receptors

dd dd

..

Delay

Receptors

TTdd

Model Model PredictivePredictiveImpedance Impedance

ControlControl

Model Model PredictivePredictiveImpedance Impedance

ControlControl

Page 9: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Example 1: Rhythmic MovementExample 1: Rhythmic MovementExample 1: Rhythmic MovementExample 1: Rhythmic Movement

Page 10: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint
Page 11: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Rhythmic Movement ErrorsRhythmic Movement ErrorsRhythmic Movement ErrorsRhythmic Movement Errors

Page 12: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Rad

.

-0.10

0.30

0.70

1.10

0.00 0.60 1.20 1.80 2.40 3.00

Ref. Model

Model Response for Rhythmic Model Response for RhythmicMovementMovement

Model Response for Rhythmic Model Response for Rhythmic MovementMovement

Time (s)Time (s)

Page 13: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

-100-50

050

100

0 0.6 1.2 1.8 2.4 3

Nm

Dist. Dist.-p Dist.-n

External DisturbancesExternal Disturbances External DisturbancesExternal Disturbances

Time (s)Time (s)

Page 14: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

-0.10

0.30

0.70

1.10

0.00 0.60 1.20 1.80 2.40 3.00

Ra

d.

Output Output1 Output2

Model Mismatch Responses Model Mismatch Responses for Rhtymic Movementfor Rhtymic Movement

Model Mismatch Responses Model Mismatch Responses for Rhtymic Movementfor Rhtymic Movement

Time (s)Time (s)

Page 15: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Example 2: Tracking MovementExample 2: Tracking MovementExample 2: Tracking MovementExample 2: Tracking Movement

Page 16: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Tracking Movement ErrorsTracking Movement ErrorsTracking Movement ErrorsTracking Movement Errors

Page 17: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Tracking MovementTracking MovementTracking MovementTracking Movement

Page 18: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

JJ 1.43 1.43 1.61 1.61 2.30 2.30 3.27 3.27BB 1.43 1.43 1.94 1.94 2.51 2.51 3.04 3.04KK 1.43 1.43 1.48 1.48 1.59 1.59 1.73 1.73TT 1.43 1.43 2.32 2.32 2.50 2.50 2.75 2.75gg 1.43 1.43 1.61 1.61 2.28 2.28 3.02 3.02

J-B-KJ-B-K 1.43 1.43 1.53 1.53 3.14 3.14 6.596.59

Errors of Parameter MismatchErrors of Parameter Mismatch( Rhythmic Movement ) ( Rhythmic Movement )

Errors of Parameter MismatchErrors of Parameter Mismatch( Rhythmic Movement ) ( Rhythmic Movement )

Parameter(s) 0% Parameter(s) 0% 15% 15% 30% 30% 45% 45%

Error is root mean square errors (rad).Error is root mean square errors (rad).

Page 19: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

JJ 0.41 0.41 0.42 0.42 0.44 0.44 0.46 0.46BB 0.41 0.41 0.43 0.43 0.45 0.45 0.47 0.47KK 0.41 0.41 0.43 0.43 0.46 0.46 0.48 0.48TT 0.41 0.41 0.42 0.42 0.43 0.43 0.44 0.44gg 0.41 0.41 0.40 0.40 0.48 0.48 0.86 0.86tdtd 0.41 0.41 0.45 0.45 0.50 0.50 0.57 0.57

J-B-KJ-B-K 0.41 0.41 0.44 0.44 0.51 0.51 0.70 0.70

Errors of Parameter MismatchErrors of Parameter Mismatch( Tracking Movement ) ( Tracking Movement )

Errors of Parameter MismatchErrors of Parameter Mismatch( Tracking Movement ) ( Tracking Movement )

Parameter(s) 0% Parameter(s) 0% 15% 15% 30% 30% 45% 45%

Error is root mean square errors (rad).Error is root mean square errors (rad).

Page 20: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

Example 3: GaitExample 3: GaitExample 3: GaitExample 3: Gait

Page 21: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint
Page 22: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

G3 is determined by linearization of the equations for small angle deviationsG3 is determined by linearization of the equations for small angle deviations

Page 23: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

X =AX+BU

Y =CX+DU

.

bS +

M P C

x0

1 2

_____________

(T1S+1)(T2S+1)

1

Step Function

Pendulum Dynamics

Dynamic Impedance PD Controller)

Angle of Ankle Joint

Identification Control

Desired Trajectory

Page 24: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint

-80

-40

0

40

80

0 0.6 1.2 1.8 2.4 3

Impulse Response Control Signal

Time (s)Time (s)

Changes of Impulse Response & Control Changes of Impulse Response & Control Signal in Double Pendulum ModelSignal in Double Pendulum Model

Changes of Impulse Response & Control Changes of Impulse Response & Control Signal in Double Pendulum ModelSignal in Double Pendulum Model

Page 25: Model Predictive Impedance Control MPIC. Motor Control Features 1.Feedback (closed loop) 2.Feedforward (open loop) 3.Learning 4.Predictive Control 5.Joint