ROBOTIC RAPID PROTOTYPING - unipi.it · ROBOTIC RAPID PROTOTYPING High Level Components ... DYNAMIC...
Transcript of ROBOTIC RAPID PROTOTYPING - unipi.it · ROBOTIC RAPID PROTOTYPING High Level Components ... DYNAMIC...
ROBOTIC RAPID PROTOTYPING
High Level Components
Manuel Bonilla, Manolo Garabini, Carlos Rosales, Alessandro Settimi e Antonio Bicchi
Tim
elin
e
• MECHANICS • Rapid Prototyping: Tecniche & Materiali
• Dallo Sketch allo Slicing
• Design for 3D rapid prototyping
• Esempi
• ELECTRONICS • Attuatori & Sensori di Base
• Schede di Controllo
• Protocolli di Comunicazione
• Esempio
• HW-IN-THE-LOOP • Comunicazione pc-hw
• Problematiche Soft REal-Time, HArd Real-Time
• MATLAB/Simulink
• Esempio: Variable Stiffness Actuators
• HIGH LEVEL COMPONENTS • “Attuatori” & “Sensori” High level
• Simulatori
• Sistemi integrazione ROS/YARP
• Video Session & Darpa Robotics Challenge
How do we connect everything?
Component drivers
ACTUATORS
• VSA*
• Robotic hands* (PISA/IIT soft hand, Velvet gripper)
• Haptic displays*
• KUKA Light Weight Robot
SENSORS
• Primesense & Kinect (RGB-D sensors)
• Phase Space (motion tracking system)
• ATI (F/T sensors)
• IMU & GPS (from tablets, mobiles, etc)
*developed at Centro "E. Piaggio"
Libraries and Tips
C++ LIBRARIES
• 3D Point clouds processing
• Computer Vision
• Grasp Analysis and Tactile Sensing
• OROCOS Kinematics and Dynamics Library
• Open Motion Planning Library
• Rigid Body Dynamic Library
• Linear Algebra
• Optimization library
• And thousands more...
Trade-off between developing time and customization!
An example - ATI and KUKA integration
Simulators – why?
READY-TO-USE
• ADAMS
• GAZEBO
• SIMULINK
DYNAMIC SOLVERS
• Rigid Body Dynamic Library
• Open Dynamic Engine
• Bullet Physic Library
Darpa Robotics Challenge
Darpa Robotics Challenge
DARPA’s Vision for Disaster Response Robots
• Disasters are unpredictable.
• Robot key properties to be effective:
• Compatibility with environments engineered for humans, even if they are
degraded.
• Ability to use a diverse assortment of tools engineered for humans.
• Ability to be supervised by humans who have had little to no robotics
training.
• Successful robots will demonstrate supervised autonomy in perception and decision-making, and be adaptable enough to operate effectively in unexpected environments where communications might be degraded.
Darpa Robotics Challenge
• Task 1 - Vehicle
• Task 2 - Obstacle
Darpa Robotics Challenge
• Task 3 – Ladder • Task 4 - Debris
Darpa Robotics Challenge
• Task 5 - Door
Darpa Robotics Challenge
• Task 6 - Wall
Darpa Robotics Challenge
• Task 7 - Valve
Darpa Robotics Challenge
• Task 8 - Hose
Darpa Robotics Challenge
How do we connect everything?
Previous tasks require:
• Sensory Perception
• Object Grasping and Manipulation
• Locomotion
• Motion Planning
• High Level Control
Middleware (Software glue)
Peer-to-peer Network of Processes using Streaming Protocols
ADVANTAGES
• Several re-usable code
• Fault tolerance
• High-level debugging tools
DISAVANTAGES
• Extra-effort when developing
• Slow (compared to hard-real time)
Integration with ROS/YARP
• Hardware abstraction
• Distributed System
• Core/Server cares about
sharing resources
– Robot Operating System
Integration with ROS/YARP
• C++
• Python
– Yet Another Robot Platform
Integration with ROS/YARP
Yarp Server • C++
Yarp Device
Yarp Port
Integration with ROS/YARP
Simple examples - Motion tracking of 3D-printed part
Integration with ROS
Complex example - ROBLOG Project Integration
What do we have to integrate?
Camera
KUKA LWR
Velvet Gripper
Grasp Planner
Motion Planner
Virtual Environment
Object Recognition
Real Time Feedback
Real Time Obstacle Avoidance
Complex example - Grasping in an Unstructured Environment