Design Optimization,

Kinematics & Dynamics

Hannes Gamper

CERN Academic Training: Robotics

 27 January 2022 / Academic Training Lectures Hannes Gamper 2

Content
1. What is Robotics?

2. Dynamics

3. Kinematics

4. Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 3

Robot
What is Robotics?
Mechanics
Motion Planning
Motion Control
Industrial Kinematics Linear, Non-Linear,
Precise Robotics Forward & Inverse Heuristic, Model Based,
Industrial Real- Localization
Robot Kinematics, parallel, Machine Learning,
& Mapping
Time PCs serial, holonom, … Primary/Secondary,
Models
Teleoperation…
Structured
Environment Model Dynamics Path
Based Lagrange, rigid, Planning
Parameter elastic Identification,
Control Inverse Dynamics, …
Identification Software Trajectory
Integration/ Planning

Embedded
Deployment
PCs Limited Cognition
AI, Machine (ROS, CRF, …) Actuation
Mobile Resources
Learning, State Motors, Drivers,
Robots Unstructured Machines, … Power Supply
Environment Sophisticated
Sophisticated Perception & Manipulators
Locomotion Perception
Locomotion Cognition Gait Analysis, Sensors - Camera, Grasping,
Sensor Stability, Redundancy, …
Lidar, Force
Integration Poincare Cycles,
…
 27 January 2022 / Academic Training Lectures Hannes Gamper 4

Robot
What is Robotics?
Mechanics
Motion Planning
Motion Control
Mechanical Robot Design Kinematics Linear, Non-Linear,
Forward & Inverse Heuristic, Model Based,
Kinematics, parallel,
Localization
Machine Learning,
Actuators serial, holonom, … & Mapping
Manufacturing Primary/Secondary,
Motors, Drivers, Teleoperation…
Power Supply
Dynamics Path
CAD Design Lagrange, rigid, Planning
elastic, Identification,
Manipulators
Inverse Dynamics, …
Grasping, In House Software Trajectory
Redundancy, … Prototyping Integration/ Planning
Deployment
External
Locomotion Cognition
Gait Analysis,
Manufacturing AI, Machine (ROS, CRF, …) Actuation
Stability, Learning, State Motors, Drivers,
Poincare Cycles, Machines, … Power Supply
…

Locomotion Perception Manipulators

Optimal Design Gait Analysis, Sensors - Camera, Grasping,
Topology, Light Stability, Lidar, Force Redundancy, …
Weight, FEA, … Poincare Cycles,
…
 27 January 2022 / Academic Training Lectures Hannes Gamper 5

Content
1. What is Robotics?

2. Dynamics

3. Kinematics

4. Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 6

Dynamics
• The equation of motion

nonholonomic
system
mechanical
solution

RP measurements on old LHC TDE w/ CERNbot

 27 January 2022 / Academic Training Lectures Hannes Gamper 7

Dynamics
• Projection equation for
more complex system

Inverse Dynamics
Trajectory Robot
= Robot -1

1
 27 January 2022 / Academic Training Lectures Hannes Gamper , David Forkel, Jorge Garai 8

1. Mass spring damper system 2. Discrete Frequency domain

Dynamics
Example: Compliant Control
3. Implementation in C++
Framework in Direct Form II

4. Controller

Low Damping: High Damping:

TIM Handling Radioactive Source for BLM Tests

Problem: Solution:
 27 January 2022 / Academic Training Lectures Hannes Gamper 9

Content
1. What is Robotics?

2. Dynamics

3. Kinematics

4. Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 10

Kinematics
Direct
Kinematics
Jacobian

Inverse
Kinematics
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions

Inverse Kinematics

Heuristic Jacobian Based

Solution for (CCD, FABRIK, (DLS, Feedback, Transpose,
Triangulation,...) …)

Solution for

Def. of Error

Final Algorithm
Other: Monte Carlo, Mesh-based, Model Predictive Control,…
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Orientation

2 Difficulties:

1. Rotations are not commutative!

2. Representation Singularities!

Solution for
Conventions:
• Euler: z -> x -> z
Solution for • Kardan: z -> y -> x (yaw – pitch – roll)

• Axis-Angle: unit vector defines axis + angle

Def. of Error
• Quaternions:

Final Algorithm
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Jacobian Matrix

Solution for

Solution for
• Analytic Jacobian • Geometric Jacobian
• Contains representation • No representation singularities!
Def. of Error singularities! • Derivative can be avoided, by
deriving the Jacobian via the
kinematic chain!
Final Algorithm
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Jacobian Inverse

Redundancy

Solution for

Solution for

Def. of Error • Moore-Penrose pseudo

inverse
• Minimizing velocities
Final Algorithm
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Jacobian Inverse

Redundancy Singularities

Solution for
Finite velocity in task
space
Solution for

Def. of Error

Final Algorithm Damped least-squares pseudo-inverse

Infinite velocity in
joint space
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Optimization Based 1st Order Redundancy Resolution

Solution for

Solution for

Def. of Error

• Minimizing joint velocities

Final Algorithm • Yields feasible solutions
• Corresponds to Moore-Penrose
pseudo-inverse
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Optimization Based 1st Order Redundancy Resolution

Artificial potential energy:

Solution for

Solution for

Def. of Error

Final Algorithm • Minimizing joint velocities

• Yields feasible solutions
• Possibility of Different Objectives
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Optimization Based 1st Order Redundancy Resolution

Solution for

Solution for

Def. of Error

Final Algorithm
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Numeric Stabilization

Integration Methods, e.g. Euler will drift

Define an Error

Solution for
Deriving the differential Equation for e
Solution for

Diff. Equation asymptotically stable iff K > 0!

Def. of Error

Final Algorithm
Leads to Closed Loop Inverse Kinematics - CLIK
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions Definition of an Error

Position

Orientation
Solution for

Solution for with

Def. of Error
• Orientation Error derived from Axis/Angle representation
• Using this error directly in feedback-loop is a
Final Algorithm simplification, but stability can be easily proven with
Ljapunov
 27 January 2022 / Academic Training Lectures Hannes Gamper

Inverse Kinematics
Conventions

Solution for

Solution for

Def. of Error

Final Algorithm
 27 January 2022 / Academic Training Lectures Hannes Gamper 22

Inverse Kinematics Kinematics Library in C++ (IK for highly redundant systems):
• Redundancy resolution for higher dexterity
• Null space projections for task priority assignment

Robotic Manipulator for FCC

Alejandro Diaz Rosales, Laura Rodrigo Perez

 27 January 2022 / Academic Training Lectures Hannes Gamper 23

Content
1. What is Robotics?

2. Dynamics

3. Kinematics

4. Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 24

Design Process

Initial Study
Requirements

Restrictions

Integration
&
Logistics

Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 25

Design Process

Initial Study
Requirements

Restrictions

Integration
&
Logistics

Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 26

Design Process

• Good Access for

Initial Study Inspections

Requirements • Allowed Area for

moving Robot
Restrictions
• Allowed Area for
permanently
installed material
Integration
&
Logistics

Design Optimization
 27 January 2022 / Academic Training Lectures Hannes Gamper 27

Design Process

Initial Study
Requirements

Restrictions

Integration
&
Logistics Requirements:
• Space: • Mass:
• Reach points I-IV (workspace of • Min. payload (~15kg)
3.35x5.50x10^5m) • Max. robot weight (~300kg)
Design Optimization • Pack up in limited space (2.9x0.55m)
while moving along tunnel axis
• Avoid Obstacles

Gamper, H.; Gattringer, H.; Müller, A. and Di Castro, M. (2021). Design Optimization of a
Manipulator for CERN’s Future Circular Collider (FCC). In Proceedings of the 18th International
Conference on Informatics in Control, Automation and Robotics, ISBN 978-989-758-522-7
 27 January 2022 / Academic Training Lectures Hannes Gamper 28

Optimization Goals:
Design Process • Min. the Degree of Freedom (DoF)
• Min. the Robot Link Lengths
State of the Art: • Min. kinematic and dynamic perf. criteria

Initial Study
Requirements

Restrictions
„Topology
Optimization“
Integration
&
Logistics

Design Optimization

Optimization of Geometry according to Performance Criteria „Geometry

Optimization“

Optimal Design
 27 January 2022 / Academic Training Lectures Hannes Gamper 29

Optimization Goals:
Design Process • Min. the Degree of Freedom (DoF)
• Min. the Robot Link Lengths
• Min. kinematic and dynamic perf. criteria

Design Space
Initial Study
Requirements
Kinematic Model
Restrictions Pruning
Continuous
Optimization Problem
Integration Pruning Function
&
Logistics Optimization
Scheme

Design Optimization
Optimal Design
 27 January 2022 / Academic Training Lectures Hannes Gamper 30

Optimization Goals:
Design Optimization • Min. the Degree of Freedom (DoF)
• Min. the Robot Link Lengths
• Min. kinematic and dynamic perf. criteria

Design Space

Kinematic Model
Pruning
Continuous
Optimization Problem
Pruning Function

Optimization
Scheme

n
Optimal Design
 27 January 2022 / Academic Training Lectures Hannes Gamper 31

Example: FCC Robot

Design Process

Initial Study
Requirements

Restrictions

Integration
&
Logistics

Design Optimization Optimal Solution:

• Optimal Geometric Parameters (link lengths)
• Optimal Topology (11 DoF, Joint Configuration)

Gamper, H.; Gattringer, H.; Müller, A. and Di Castro, M. (2021). Design Optimization of a
Manipulator for CERN’s Future Circular Collider (FCC). In Proceedings of the 18th International
Conference on Informatics in Control, Automation and Robotics, ISBN 978-989-758-522-7
 27 January 2022 / Academic Training Lectures Hannes Gamper 32

Design Process Gamper, H.; Luthi, A.; Gattringer, H.; Müller, A. and Di Castro, M.;
Design Optimization of Quality Inspection Robots for Particle Accelerator
Components, In Proceedings of the ECCOMAS Multibody Dynamics
Conference, 2021
Example: Cavity Design Optimization

Initial Study • Visual inspection of inner surface

Requirements after assembly
• Small allowed robot space
• Big workspace for inspection
Restrictions
• No restrictions on robot design
(topology, geometry)

Integration
& FCC LINAC LHC
Logistics
Tentative
Optimized

Design Optimization

FCC LINAC LHC

 27 January 2022 / Academic Training Lectures Hannes Gamper 33

Design Process Gamper, H.; Luthi, A.; Gattringer, H.; Müller, A. and Di Castro, M.;
Design Optimization of Quality Inspection Robots for Particle Accelerator
Components, In Proceedings of the ECCOMAS Multibody Dynamics
Conference, 2021
Example: Cavity Design Optimization

Initial Study
Requirements

Restrictions

Integration
&
Logistics

Design Optimization
27 January 2022 / Academic Training Lectures Hannes Gamper 3
4

Thank you
for your attention!