Robotics: Basics and Advanced Concepts

Application of Parallel Robots

Ashitava Ghosal
Professor
Department of Mechanical Engineering
&
Centre for Product Design and Manufacturing
Indian Institute of Science, Bangalore

NPTEL, 2020
 Contents
◼ Lecture 1
Sun Tracking using a 3-DOF parallel manipulator
◼ Lecture 2
Stewart-Gough platform based force-torque sensor
◼ Lecture 3
Vibration isolation using a Stewart-Gough platform
 Robotics & Design Lab
@IISc

◼ Acknowledgements
Ashith Shyam R B
Mohit Acharya
Nils Bjorkman
Sudipta Pramanik
…. several other students
◼ Funding by SERIIUS
 Contents – Lecture 1
◼ Introduction
◼ Sun tracking – Az-El and T-A configurations
◼ Sun tracking using redundant parallel robot
→ 3-RPS parallel robot
→ 3-UPU parallel robot
◼ Conclusion
Introduction
Solar Energy Harvesting

Flat photovoltaic panels

• ~18 – 20% efficiency
• Storage required
• Competitive with coal
• Kurnool, India – 1GW

Parabolic trough
• Rotation about one axis
• Incident energy focused
on to a tube at focus
• ~20 % efficiency

Dish concentrator
• Parabolic dish
• Incident energy concentrated on to a
Stirling engine
Source: Google images/Wikipedia • Complete assembly tracks the sun – 2 axis motion
• Efficiency larger than PV/Parabolic trough >30%
 Concentrated Solar Power Tower

Spain, USA, Australia, Middle East, North Africa, India, China etc.
◼ Concentrated solar power
→ large number (> 5000) of mirrors/heliostats
→ track sun and focus energy at a distant receiver (~500 m to ~1 km)
-- high temperature at receiver → higher efficiency ~30%
-- store solar energy as heat → no need for batteries
→ Very slow motion of the sun ~ 180 degrees in 12 hours; 2 axis motion
→ High accuracy requirement ~ 5 mrad pointing accuracy
 Solar Power Tower Technology

◼ Heliostat cost
◼ ~50 % of total investment cost
(Kolb, 2009) – Target 80$ /m2
◼ Solar Power Tower technology
shows high potential
for technology improvements
(Lovegrove, 2012)

Picture source: Kolb (2007)

 Existing Heliostats

Location of the sun in the sky known

Incident ray (depends on latitude, longitude,
Reflected ray time of day and date )
→ Incident ray known
Mirror normal
Location of receiver known
→ Reflected ray known

Two well known existing methods

• Azimuth-Elevation (Az-EL)
• Target-Aligned (T-A) or Spinning
Elevation

Rotations about two perpendicular axes

Mirror to align mirror normal appropriately
 Limitations of Existing Heliostats

• Deflection may exceed slope error budget of 2-3 mrad due to

wind loading and/or self-weight
→ Operational wind loading up to 10 m/sec, Survival wind loading 22 m/s

• Need large amount of backing/support material for avoiding deflection

• Need accurate and expensive speed reducers to achieve ~15 degrees

per hour
Parallel mechanism for Sun Tracking
◼ Many advantages as a parallel
manipulator
→ More load carrying capacity
3-RPS Parallel Manipulator -- can carry larger mirrors
-- less deflection and hence less use of
structural supporting material
→ More accuracy
◼ Can use linear actuators – no need
for expensive speed reducers
◼ One extra actuator → redundant
3 principal motion – rot(X),rot(Y)
and tran(Z)
◼ Redundant system → Constraint
Keep Zc constant

Ashith Shyam R B
3-RPS Manipulator (Contd.)
Kinematics of 3-RPS
• Given location of sun in the sky, obtain
a) Orientation of plane of top platform (mirror)
b) Translations of prismatic joints to achieve the orientation

Algorithm
1) Choose zc arbitrarily
2) From laws of reflection

3) Orientation of top platform with respect to bottom plate -- [R] = [n o a] with

CJME Ashith Shyam R B

 3-RPS Manipulator (Contd.)
Algorithm (Contd.)
4) 3-RPS configurations introduces 3 additional constraint

where rp is the distance from G to Si

5) 8 equations in 8 unknowns – Use Bezout’s elimination to arrive at two equations in two unknowns –
6) Solve numerically for
7) Obtain [n o a] (rotation matrix) from

8) From rotation matrix obtain location of S1, S2 and S3 with respect to bottom platform

9) Obtain leg lengths as

CJME Ashith Shyam R B

3-RPS Manipulator – Simulation results

Simulation results for Bangalore, India

May 6, 2013 CJME -- Ashith Shyam R B
3-RPS Manipulator – Simulation results

Simulation
done in Matlab

Ashith Shyam R B
 Sun Tracking – 3-RPS parallel

Ashith Shyam R B
 FEA Analysis
• CAD Model in SolidWorks

• FEA in ANSYS Work bench

• Loading – Self-weight
(10 kg/m2)
+
Wind load of 22 m/s

• Worst case analysis

Ashith Shyam R B - CJME

 Sun Tracking – Hardware
videos

Azimuth-Elevation Configuration

Ashith & Mohit

 Sun Tracking – Hardware

3-RPS Configuration – Fabricated

Ashith & Mohit

Experimental Results

Ashith Shyam R B – Solar Energy

Experimental Results (Contd.)

• Sun tracking using Azimuth-Elevation & 3-RPS Heliostat

• Pointing errors larger for 3-RPS ~30 mrad vs ~20 mrad

• Main cause – manufacturing & initial settings

Ashith Shyam R B -- Solar Energy
 Parallel mechanism for Sun
Tracking – 3-UPU Manipulator
• Three actuators – Prismatic (P) joints & 2 Universal (U) joints in each leg → 3 DOF according to Grubler-Kutzbach criterion

• Special axes geometry

-- Two intermediate R axis are parallel, first and last R axis intersect at a point -- Stationary centre of rotation

• Spherical ``wrist’’ – has three rotational degrees of freedom

Karouia & Herve, ARK (2000), R Di Gregorio, MMT (2003)

3-UPU Sun Tracker

3-UPU Parallel Manipulator Kinematics

From the location of the sun
& receiver -- normal to mirror

Orientation of mirror M with

respect to base B

Note:

Obtain the other two columns of [R] for Az-El and T-A modes
 3-UPU Sun Tracker
3-UPU Parallel Manipulator Kinematics (Contd)
Azimuth-Elevation – two consecutive Euler angles
a) Rotation about Z by azimuth angle

b) Rotation about ym by angle where elevation angle is

Rotation matrix for Az-El configuration

Since the last column is known, azimuth and elevation angle can be obtained

Ashith Shyam R B MMT

 3-UPU Sun Tracker
3-UPU Parallel Manipulator Kinematics (Contd.)
◼ Rotation matrix for Target-Aligned (T-A) configuration

◼ Two consecutive Euler angles

a) Spin about

b) Rotate about ym by elevation angle

◼ Rotation matrix for Spinning-Elevation (T-A)

◼ Comparing last column obtain the two angles

Ashith Shyam R B MMT
 3-UPU Sun Tracker
3-UPU Parallel Manipulator Kinematics (Contd.) -- Actuation to achieve orientation of mirror

◼ Location of U joint in bottom platform

◼ Location of U joint in the top platform

◼ Position vector of top U joint from rotation matrix [R] and translation
vector

◼ Obtain leg lengths from

Ashith Shyam R B MMT
 3-UPU Sun Tracker
3-UPU Parallel Manipulator Kinematics (Contd.)
Actuation required to achieve orientation of mirror
March Equinox at Bangalore, India

Leg lengths in T-A mode

Configuration of 3-UPU in T-A mode

Ashith Shyam R B MMT
 3-UPU Sun Tracker
3-UPU Parallel Manipulator Kinematics (Contd.)
Actuation required to achieve orientation of mirror
March Equinox at Bangalore, India

Leg lengths in Az-El mode

Configuration of 3-UPU in Az-El mode

Ashith Shyam R B MMT
 Sun Tracking using Az-EL & T-A
Spillage losses – image of plane mirror on a receiver during March Equinox

(a) 9 am (b) 12 noon (c) 2 pm (d) 5 pm

March Equinox
December Solstice

Ashith Shyam R B MMT

 3-UPU Sun Tracker
3-UPU Parallel Manipulator

CAD model of 3-UPU ``wrist’’ and leg details

Centre of rotation is above the plane of the mirror

Sun Tracking using 3-UPU

• Switched between for Az-El or Target-Aligned configurations

• e-Reconfigurable – no change in hardware
• Experiments done at IISc
Ashith Shyam R B MMT
 3-UPU as TA and Az-EL

Bangalore @ roof of ICER

May 24, 2017

Receiver centre at
(0, 0 6.72) m

Mirror centre at
(-10, 3, 0) m

Az-El 3:18 pm T-A 3:20 pm

Ashith Shyam R B MMT
 Conclusion
◼ Novel use of a parallel manipulator for sun tracking
◼ Kinematics & Design Challenges
◼ Several advantages
→ More load carrying capacity
→ Use of low-cost linear actuators
→ Potentially more accurate
→ e-Reconfigurable – use of redundant actuation in 3-UPU
◼ Accuracy needs to be improved by
→ Removing play at the joints
→ Better manufacturing
Robotics: Basics and Advanced Concepts
Application of Parallel Robots

Ashitava Ghosal
Dept. of Mechanical Engineering
&
Centre For Product Design and Manufacture
IISc Bangalore
NPTEL 2020
Acknowledgement

• Dr. R. Ranganath @ ISAC-ISRO

• Dr. Sandipan Bandyopadhyay @IIT-M
Contents – Lecture 2

▪ Introduction
▪ Statics of a 6– 6 Stewart-Gough Platform
▪ Singular Configurations in a 6-6 Stewart Platform
▪ Design of 6 component force-torque sensor
▪ Experiments
▪ Conclusion
 Introduction

Gough- Stewart Platform

→ First used as a tire-testing machine in UK

→ Six actuated extendable legs -- 6 DOF

Linear motion along X, Y and Z

Rotational motion about X, Y and Z

Also known as Heave, Surge, Sway

&
Roll, Pitch and Yaw
Stewart 1965 Extendable `legs’
 Modern uses of Gough - Stewart platforms

Industrial – material handling Micro-motion

Robotic surgery

Industrial – manufacturing Precise alignment of

mirrors
Gough-Stewart Platform

Cabin with audio-visual

display

Flight Simulators
Legs of Gough-Stewart platform
Gough- Stewart
Platform
→Moving Top Platform
→Fixed Base

6 actuated joints

→Actuating joints
in a coordinated
manner results
in motion of top
platform
Motion Simulation

Simulations
done using
ADAMS

Individual motions and combined motion

of a Gough-Stewart platform -- simulations
Santosh @ APDAP
 Gough-Stewart Platform as a Sensor

• Gough-Stewart platform as a 6 –axis force torque sensor

→ Actuators are locked – 0 degrees of freedom
→ Instead of applying force at actuators/cylinders, strain gauge
based sensors put there to measure strains
→ Externally applied force-moment at the top platform can be
sensed at the legs
 Statics of Gough-Stewart platform

 Unit vector along leg

External load – force & moment

 Leg Forces

A leg of a Stewart platform

Final formula → f = inv ([H])(F;M)
 6 Component Force-Torque
Sensor
• Basic formula f = inv ([H])(F;M)

→ Matrix [H] depends on chosen geometry

→ (F;M) -- 6 components of externally applied force and moments
→ f – axial forces in the 6 legs
• If [H] is isotropic, all components are equally sensitive
• If [H] is singular, certain components will be amplified ``mechanically’’
• Can compute directions where amplification occurs
Singularities

Singularity manifold at a
given orientation of top
platform
Bandyopadhyay & AG MMT 06
Stewart Platform configuration

• Assume – top and bottom platform same size

• Bottom platform points are Bi, i=1,…,6

• Top platform points are Pi, i=1,…,6

• Leg connections bi pj determine configurations

• Obtain singular directions of from [H] for
different configurations
Singular directions of Stewart Platform configurations
SL Leg connections ‘0’ Singular
No
Leg 1 Leg 2 Leg 3 Leg 4 Leg 5 Leg Eig.Val Directions
6 of [H]
1 b1-p1 b2-p2 b3-p3 b4-p4 b5-p5 b6-p6 3 Fx, Fy, Mz

2 b1-p2 b2-p1 b3-p3 b4-p4 b5-p5 b6-p6 2 Fx, Mz

3 b1-p2 b2-p1 b3-p4 b4-p3 b5-p5 b6-p6 1 Mz

4 b1-p2 b2-p1 b3-p4 b4-p3 b5-p6 b6-p5 0 None

5 b1-p3 b2-p2 b3-p1 b4-p4 b5-p5 b6-p6 2 [1/3,1,0]1, Mz,,

6 b1-p4 b2-p2 b3-p3 b4-p1 b5-p5 b6-p6 2 Fy, Mz,

7 b1-p1 b2-p3 b3-p2 b4-p4 b5-p6 b6-p5 1 Fy

8 b1-p2 b2-p3 b3-p4 b4-p5 b5-p6 b6-p1 2 Mx My Mz

Note: Superscript ‘1’ denotes singular force direction in a

plane.
14
6 Component Force-Torque Sensor

Flexible hinges (instead of joints) to overcome

friction & hysterisis
Ranganath et al. MMT 04
• Prototype 1 -- ISRO
Force sensing
FEA in NISA ring element in leg

Prototype 1

Ranganath et al. MMT 04

 Sample
experimental
results

Shows enhanced
sensitivity for Fx,
and Fy components

Ranganath et al. MMT 04

 Sample
experimental
results

Shows enhanced
sensitivity for Mz
component

Ranganath et al. MMT 04

 Test Results & Comparison
Wrench reconstruction from test 1
Particulars W WFEA WTEST
(Fx+Fy) acting together Fx +0.093 0.0942 0.0945
Fy +0.093 0.0898 0.0899
Fz 0 0.0095 0.0027
Mx -1.0230 -1.9576 -1.3260
My +1.0230 1.2617 1.0376
Mz 0 -0.0536 -0.1463

Load Applied FEM Test

Fx 0.0930 0.0895 0.0911
Fy 0.0930 0.0944 0.0933
Fx, Fy and Mz acting Fz 0 0.0274 0.0149
together Mx -1.0230 -2.2302 -1.245
My +1.0230 1.8440 0.9173
Mz -4.6500 -4.4592 -4.6847
• Prototype 2 – sensitive to Mx, My & Mz

CAD model of a leg

with sensing ring

Flexible hinge –
2 DOF
CAD model of sensor
Prototype 2
Six component force-torque sensor with
enhanced sensitivity to moments.

Measure forces up to 50 N in X and Y

and 200 N in Z direction with sensitivity of 0.5N

Measure moments up to10,000 N-mm with

sensitivity of 50 N-mm

Stewart in a near-singular configuration with

flexible hinges

Experimental test results

for My

Plot of strain (leg1) vs My

http://www.mecheng.iisc.ernet.in/~asitava/NaCoMM-2007-061.pdf
Conclusion
• Gough - Stewart platform, a versatile device
– used in wide range of tasks.

• Extensively used as motion platform

– flight simulators in aerospace industry.

• Gough - Stewart platform as 6 component force-torque sensor.

• Statics equations & singular configurations.

• Two prototypes, calibration & testing.

Robotics: Basics and Advanced Concepts
Application of Parallel Robots

Ashitava Ghosal
Dept. of Mechanical Engineering
&
Centre For Product Design and Manufacture
IISc Bangalore
NPTEL 2020
Acknowledgement
Nazeer Ahmad – Ph D student from ISAC-ISRO
Contents – Lecture 3
• Motivation
• Stewart Platform Modeling for Vibration Isolation
• Simulation Results
• Experimental results
• Conclusion
Motivation
❖ On-orbit vibrations

DGA ✓ optical performance degrades

✓ Inter-satellite link communication
PAA
❖ On-orbit vibration sources
RW ✓ Appendage Deployment
✓ Thruster Firing, LAM firing
✓ Momentum Wheel/ RW
✓ Pump operation
 Problem definition: On-orbit isolation

Source characteristics
❖ Multiple multi-axis noise sources
❖ 1000 µg RMS with peak value of 3000 µg spread over a
bandwidth of up to 250 Hz.
Isolation system requirements
❖ A multi-axis Vibration Isolation System is required
❖ Isolation for all DOFs – X, Y, Z and Rot(X), Rot(Y) and Rot(Z)
❖ The first 6 frequencies of the isolator should be decoupled
and in a narrow bandwidth – ideally all equal or unity
dynamic isotropy
❖ Six axis Stewart platform can provide isolation in 6
directions and thus is initial natural choice
 Stewart-Gough platform (SGP) based multi-axis
vibration isolation system for onboard isolation
❖ Equation of motion of SGP in terms of 5 parameters 𝑅𝑎 , 𝑅𝑏 , 𝛼, 𝛽, 𝐻

𝐌 𝛘 𝛘ሷ + 𝐟𝐜 𝛘, 𝛘ሶ + 𝐂𝛘ሶ + 𝐆 𝛘 + 𝐊𝛘 = 𝐟

❖ Eigenvalue problem: 𝐊 − 𝐌𝜔2 =0

Assumptions for eigenvalue problem: Base fixed, frictionless SPS configuration

❖ Equivalent model in MSC ADAMS developed

Optimization Problem
𝝎𝒎𝒂𝒙
𝐦𝐢𝐧𝐢𝐦𝐢𝐳𝐞: 𝜿 =
𝝎𝒎𝒊𝒏

Subject to: 100 𝑚𝑚 ≤ 𝑅𝑎 , 𝑅𝑏 ≤ 300 𝑚𝑚

0𝑜 ≤ 𝛼, 𝛽 ≤ 120𝑜

0 𝑚𝑚 ≤ 𝐻 ≤ 150 𝑚𝑚

𝑚𝑝𝑙𝑑 = 𝐶1

𝐈 𝑝 = 𝐂2
𝑝
𝐫𝐶𝐺 = 𝐂𝟑

𝐾𝑙 = 𝐶4
𝑝
𝑚𝑝𝑙𝑑 , 𝐈 𝑝 and 𝐫𝐶𝐺 is fixed by payload
𝐾𝑙 is fixed by required frequency
Simulation Results

Converged results:
𝑝
Table 2.1: 𝐫𝐶𝐺 = 𝟎 ; DII 𝜅 = 1.7
𝑝
Table 2.1: 𝐫𝐶𝐺 = 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒; DII
𝜅 = 1.6

𝑧𝑐𝑔 : −18 mm to −53 mm

 Prototype Stewart-Gough Platform
Prototype descriptions
❖ flexural joint at ends
❖ Stiffness: circular plate spring
Characterization test
❖ Base fixed modal survey tests
❖ A constant acceleration sine sweep input signal of
amplitude 0.5g used
❖ Test done in lateral and longitudinal directions.
❖ Tri-axial accelerometers mounted at the center of top and
Experimental hardware bottom platforms.
❖ FRF obtained between top and bottom platform
acceleration responses
❖ Magnitude of cross FRFs found to be very low
Comparison of computed and experimental
FRFs

Isolation of 19.1dB/octave achieved

Ahmad -- Spacecraft Technology

Conclusion
• 6 axis Stewart- Gough platform can be used for vibration isolation

• Proposed for vibration isolation of payload in spacecraft.

• Ideally all eigenvalues should be same – allow all modes to be equally well
damped and isolated.

• Work is continuing
Suggested Additional Reading – Week -5
• S. Bandyopadhyay and A. Ghosal 2009, “An algebraic formulation of static isotropy and
design of statically isotropic 6-6 Stewart platform manipulators”, Mechanism and Machine
Theory, Vol. 44, pp. 1360-1370.

• R. Ranganath, P. S. Nair, T. S. Mruthyunjaya and A. Ghosal 2004, “A force-torque sensor

based on a Stewart Platform in a near singular configuration”, Mechanism and Machine
Theory, Vol. 39, pp. 971-998.

• R. B. Ashith Shyam and A. Ghosal 2015, “Three-degree-of-freedom parallel manipulator to

track the Sun for concentrated solar power systems”, Chinese Journal of Mechanical Engi-
neering, Vol. 28, No. 4, pp. 793-800.

• R. B. Ashith Shyam, M. Acharya and A. Ghosal 2017, “A heliostat based on a three degree-
of-freedom parallel manipulator”, Solar Energy, Vol. 157, pp. 672-686, 2017.

• R. B. Ashith Shyam and A. Ghosal 2018, “Path planning of a 3-UPU wrist parallel manip-
ulator for sun tracking in central receiver tower systems”, Mechanism and Machine Theory,
Vol. 119, pp. 130-141.

• R. B. Ashith Shyam, Nazeer Ahmad, R. Ranganath and A. Ghosal 2019, “Design of a dy-
namically isotropic Stewart-Gough platform for passive micro-vibration isolation in spacecraft
using optimization, Journal of Spacecraft Technology, Vol. 30, No. 2, pp. 01-08.

The above papers contains details of research reported in this week lectures. There are several
important references contained in the papers listed above and anyone interested in the topics of
this week should take a look at them.