Matlab EXPO 2021
Matlab EXPO 2021
Perception to Motion
YJ Lim Ronal George
0
Autonomous Robot Development with MATLAB & Simulink
1
Autonomous Robot Development with MATLAB & Simulink
2
Industry Trends – Advanced Robotics
Factories are:
• Incorporating more advanced systems
• Taking advantage of new technology
• Increasing connectivity between components
• Enhancing infrastructure to reach Industrial
IoT
3
Smart Factory Concept: Asset, Operation, & Workforce Optimization
IoT
Factory Cloud + AI
Local 5G Operation
Big Data
Process Management・Quality Increase Management
Robot Optimization
✔?
✘?
OT
Worker AMR
Optimization
Autonomy Monitor
Update
4
Smart Factory Concept: Asset, Operation, & Workforce Optimization
What we are talking about today
IoT
Factory Cloud + AI
Local 5G Operation
Big Data
Process Management・Quality Increase Management
Robot Optimization
✔?
✘?
OT
Worker AMR
Optimization
Autonomy Monitor
Update
5
Robotics in Smart Factory: Trends and Directions
Conventional
▪ Automated Systems
▪ Repetitive tasks
▪ Manual programming
▪ Safety fence needed
6
Robotics in Smart Factory: Trends and Directions
Conventional
▪ Automated Systems
▪ Repetitive tasks
▪ Manual programming
▪ Safety fence needed
7
Robotics in Smart Factory: Trends and Directions
Conventional Current
✔?
✘?
8
Robotics in Smart Factory: Trends and Directions
✔?
Conventional
✘? Current
✔?
✘?
10
Robotics in Smart Factory: Trends and Directions
11
Challenges: Growing Complexity
Environment
Robot Model
Development Environment
Design Simulate Analyze Implement Test
13
Autonomous Robot Development with MATLAB & Simulink
Environment
Robot Model
Development Environment
Design Simulate Analyze Implement Test
14
Physical & Kinematic Modeling
CAD Model
Multibody
Model
Creating a physical model Automatic import from CAD Tools URDF / Robot Library
%% Import robot from URDF
smimport('GEN3_FOR_URDF_ARM_V11_MOD.urdf');
CAD Model
Multibody
Model
%% Use robot library
Robot = loadrobot('abbYuMi’);
Show(robot);
Environment
Robot Model
Rigid Body
Low-Fidelity Multibody Co-Sim with Co-Sim with
Kinematics &
Model Modeling Gazebo Unreal
Dynamics
18
Rigid Body
Fast & Low-Fidelity Simulation Low-Fidelity
Model
Kinematics
& Dynamics
Multibody
Modeling
Co-Sim with
Gazebo
Co-Sim with
Unreal
Motion Models
Manipulator dynamics model that accepts joint torques and gripper commands
System Model
Upgrade
Robotics System
Toolbox
Physical dynamics system with built-in joint limits and contact modeling
System Model
Upgrade
Simscape Multibody
Environment
Robot Model
Development Environment
Design Simulate Analyze Implement Test
25
How is deep learning used for robotics?
❑Speech Recognition
Audio Deep Learning Text
Voice Command to
Robots
26
How is deep learning used for robotics?
❑Speech Recognition
Audio Deep Learning Text Voice Command to
Robots
❑Computer Vision
Pixels Deep Learning Labels
27
How is deep learning used for robotics?
❑Speech Recognition
Audio Deep Learning Text Voice Command to
Robots
❑Computer Vision
Labels Object Identification for
Pixels Deep Learning
Robots
Image
Pixels Deep Learning
Abnormal
Industrial Inspection
with Robots / UAVs
28
How is deep learning used for robotics?
❑Speech Recognition
Audio Deep Learning Text Voice Command to
Robots
❑Computer Vision
Labels Object Identification for
Pixels Deep Learning
Robots
Pose Estimation
Localization
for Robots
❑Point Cloud Processing
Point
Deep Learning Structure
Clouds
29
How is deep learning used for robotics?
❑Speech Recognition
Audio Deep Learning Text Voice Command to
Robots
❑Computer Vision
Labels Object Identification for
Pixels Deep Learning
Robots
•
•
•
•
•
•
Automated Labeling &
Iterative Learning
•
•
•
Automated Labeling &
Iterative Learning
34
Computer Vision Toolbox, Deep Learning Toolbox, Robotics System Toolbox, Stateflow 34
Full Workflow of Pick-and-Place Robots
Robotics System Toolbox, Computer Vision Toolbox, ROS Toolbox, Image Processing Toolbox, Stateflow 35
Full Workflow of Pick-and-Place Robots
• Dynamic environment
• Flexible operations
Scan‒and‒Build Environment
36
Full Workflow of Pick-and-Place Robots
• Dynamic environment
• Flexible operations
Scan‒and‒Build Environment
1 2
37
Full Workflow of Pick-and-Place Robots
Shipping examples
• Dynamic environment
• Flexible operations
Scan‒and‒Build Environment
1 2 3
38
Full Workflow of Pick-and-Place Robots
1 2 3
39
Full Workflow of Pick-and-Place Robots
1 2 3
40
Full Workflow of Pick-and-Place Robots
41
42
Advanced Control: Reinforcement Learning
Train robot to reach the target object
AGENT
ACTION
STATE
Policy
Train to reach the target object
Policy update
Reinforcement
Learning
Algorithm
REWARD
ENVIRONMENT
AGENT
ACTION
STATE
Policy
Train to reach the target object Train to grasp the target object
Policy update
Reinforcement
Learning
Algorithm
REWARD
ENVIRONMENT
Environment
Robot Model
Development Environment
Design Simulate Analyze Implement Test
45
Hardware Connectivity & Deployment
Middleware
MATLAB API
Environment
Robot Model
ROS Driver
Environment
Robot Model
Development Environment
Simulink
50
Autonomous Robot Development for Smart Factories
✔?
✘?
Simulink
52
Use Cases: Delta Robot for Automated Parts Sorting
Connect
Sense
Perceive
Plan &
Control
Platform Decide
53
Key Takeaways
54
Learn More
55
Thank you
Contact:
© 2021 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc. See mathworks.com/trademarks
for a list of additional trademarks. Other product or brand names may be trademarks or registered trademarks of their respective holders.
56