erf2015 workshop flexibility and dexterity in industrial robots...

Post on 30-May-2020

2 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Flexibility and dexterity in industrial robots:

Demonstrators of new frontiers in industrial

applications

12th March 2015, Vienna, Austria

Compliance robot control for enhanced

industrial robots dexterity and flexibility

D. Surdilovic, Fraunhofer – IPK Berlin

Laboratory for RobotSupportedCooperativeWork

ERF2015 WORKSHOP

Objectives – Nowadays, Future Industrial Robots

Dextereous Manipulation and Flexibile Automation for Manufacturing

Applications

Ambituous, nevertheless logical demands:

Dextereous Manipulation : Area of robotics in whichmultiple manipulators or fingers cooperate to grasp andmanipulate objects;

Flexible Automation : Ability for a robot or system to be quickly and easily re-tasked to change product design for both low and high mix manufacturing (capacity-volume and product variants flexibility)

Robots are currently incapable of competing with human toperform dexterous tasks (contact tasks) in a flexible manner!

Role of compliance (dexterity ↔ flexibility)?

Active compliance control – a 30 years old idea (Mason

1985), still not alive in industrial (conventional) robotics

INTERACTING WITH A GODZILLA

WHICH SENSE IS CRUCIAL FOR CONTACT?

Pioneering work in force and compliance control (since 1987)

Challenges: sampling time, quantization, time lags….

LABORATORY FOR RSCW AT IPK

Actual HRI summary: Everybody moves and guides a robot.

Research on Active Compliance Control (30 years after) :Position based impedance and force control

Target Impedance/Force Position based impedance/force control C-frame

Robust control design

Matlab Toolbox

CartPose cl = hl.getPose();

lArm. setComplianceGains(ENGAGE_CONFIG);

lArm.setComplianceFrame(CartPose())

lArm << MoveLin(cl); CURL++ (ROS)

Open-Control (C5Gopen,

KUKA RSI , ORiN etc.)

Planing and programming

For Compliance Control Testing (including process forces

– modelling/experiments)

AUGMENTED REALITY

CLASSICAL TASKS: GRASPING AND INSERTING

Combined Force/Impedance Control

CRITICAL EDGE FOLLOWING

Things that should never be done!

Improperly designed

controller may be

destabilized by human

Maximum target

bandwidth < one half of

the position control

bandwith

DEXTEREOUS CONTACT TASK – N40 COCKPIT SET

• Several latching mechanisms

• Complex hand motion

• Compliance, motion and

force control - key strategy

• Variation of task execution

• Difficult (impossible)

conventional robot

programming and execution

• Limitations of industrial add-on

tools (e.g. active and passive

compliant devices)

LIFE CYCLE TESTING

CONSIDERABLY SLOWER ROBOT

EXECUTION (SAME FORCE LEVEL)

• Considerable amount of damping

needed to stabilize the interaction

• Relatively slow motion to keep

smaller interaction forces and torques

• Human task: Variable impedance,

rapid change of strategies

• Robotic execution considerably

slower (30 s cycle) vs. human

operation (app. 5 s)

• Development of human-like

strategy

• Action termination based on

interaction force/torque monitoring

Human-task

Improved robotic testing

Dexterous Advanced Robot

Programming

Planning and Programming – main

bottleneck for wider applications of

dual-arm robots in industry

Bi-Manual human operations –

Central interference hypothesis -

content-dependence (Hazeltine 2006) -

Bottlneck is in the planning of the

movement (motor preparation)

Mimic human motion

Simple planning and

programming of human-like

bimanual motion („Callosum“ –

control)

-Symetric/asymetric,

congruent/non-congruent

motions

-Arms collision monitoring and

avoidance

EFFICIENT DUAL-ARM INSERTION (INTERACTION BETWEEN TWO COMPLIANT ARMS)

BI-INSERT(SUBJECT& HL, SUBJECT& HR, DOUBLE SPD=INSERT)CURL++ (ROS PACKAGE)

biInsert is resolved to:

CartPose cl = hl.getPose();

lArm. setComplianceGains(ENGAGE_CONFIG);

lArm.setComplianceFrame(CartPose())

lArm << MoveLin(cl);

CartPose cr = hr.getPose();

lArm. setComplianceGains(ENGAGE_CONFIG);

rArm.setComplianceFrame(CartPose())

rArm << MoveLin(cr);

CartPose cl = hl.getPose();

lArm. setComplianceGains(INSERT_CONFIG);

lArm.setComplianceFrame(CartPose() +Vector3(0,0,-d/2)))

lArm << MoveLin(cl);

CartPose cr = hr.getPose();

lArm. setComplianceGains(INSERT_CONFIG);

rArm.setComplianceFrame(CartPose()+Vector3(0,0,-d/2))

rArm << MoveLin(cr);

lArm.execute();

rArm.execute();

Engage

Insert

FLEXIBLE AUTOMATION + DEXTEREOUS ARMS MANIPULATION – GEAR MESHING

INSTEAD OF CONCLUSION : COMPLIANCE ROBOT ARMS CONTROL -KEY TECHNOLOGY TOWARDS ENHANCED DEXTERITY AND FLEXIBILITY

top related