workshop italy-japan
TRANSCRIPT
Floating-base manipulatorstoward underwater & aerial autonomy
Floating-base manipulatorstoward underwater & aerial autonomy
Gianluca Antonelli (and many others...)
University of Cassino and Southern Laziohttp://www.eng.docente.unicas.it/gianluca_antonelli
Italy-Japan WorkshopRobot co-worker - Towards a society supported by robot co-workers
Tokyo, 11 December 2014
Gianluca Antonelli Tokyo, 11.12.2014
Outline
motivation
multiple tasks, why and how?
a possible solution
simulation & experiments
(tenth talk of the day, I known,no equations nor Lyapunov
analysis...)
(tenth talk of the day, I known,no equations nor Lyapunov
analysis...)
Gianluca Antonelli Tokyo, 11.12.2014
Aerial Robotics Cooperative Assembly System
http://www.arcasproject.eu
2011-2015
Gianluca Antonelli Tokyo, 11.12.2014
AErial RObotic system integrating multiple ARMS and advanced manipulation capabilities forinspection and maintenance
Follow up of the ARCAS project:
H2020 - call1
2015-2019
Gianluca Antonelli Tokyo, 11.12.2014
Marine AutonomousRobotics for InterventionS
In case you were wondering where Cassino is...
2012-2015
Gianluca Antonelli Tokyo, 11.12.2014
Effective Dexterous ROV Operations in Presence of Communications Latencies
2015-2018
Demo at 1000m
Gianluca Antonelli Tokyo, 11.12.2014
Redundancy
Our robotic system is plenty of DOFs
Main control objective mainly end-effector related
Let us try to keep it simple first:
● Only end-effector position/orientation(we will keep this discussion at
the kinematic level)
Gianluca Antonelli Tokyo, 11.12.2014
Redundancy
Controlledvariable
System configuration Differential
relantionship
Gianluca Antonelli Tokyo, 11.12.2014
Redundancy
classicalpseudoinerse
Gianluca Antonelli Tokyo, 11.12.2014
UW example: naïve solution
MARIS model: 6DOF vehicle + 7DOF arm
End-effector configuration only (6DOF)
Reach a pre-grasp configuration
● vehicle and arm uncoordinated
● Kinematic singularity
● Mechanical joint limits
● Roll-pitch not null
● Redundancy not exploited
Poor performance!
Gianluca Antonelli Tokyo, 11.12.2014
UW example: naïve solution
End-effector configuration only, weighted pseudoinverse
Relative weight between vehicle and arm, alone, is not the solution
Gianluca Antonelli Tokyo, 11.12.2014
Exploiting redundancy
Let us add other control objectives
Projections now need care
Gianluca Antonelli Tokyo, 11.12.2014
Multiple tasks
An almost 40-years long journey
Liégois, 1977 (null-space-gradient-based)
Maciejewski & Klein, 1985 (2-task-priority)
Nakamura, Hanafusa, Yoshikawa, 1987 (2-task-priority)
Siciliano, Slotine, 1991 (multiple-task-priority)
Chiaverini, 1997 (2-task-singularity-robust priority)
Mansard, Chaumette, 2007 (multiple-task-singularity-robust priority)
Antonelli, Arrichiello, Chiaverini, 2008 (multiple-task-singularity-robust priority)
Gianluca Antonelli Tokyo, 11.12.2014
What tasks?● End-effector
● End-effector directional sensor
● Arm manipulability
● Obstacle avoidance
● Self-hitting avoidance
● Mechanical joint limits
● Energy-related tasks
● Vehicle roll and pitch
● Vehicle yaw
● Vehicle-arm centre-of-masses alignment
● Humanoids (head, balance, etc...)And in real-time, of course...
Gianluca Antonelli Tokyo, 11.12.2014
NSB
Null-Space based Behavioral control, i.e., a behavioral interpretation
Each task inverted as if it were alone
Lower priority projected intothe null-space of combined higher priority
Gianluca Antonelli Tokyo, 11.12.2014
(supposed) features
Tasks priority guaranteed
Impedance control?
Robust to algorithmic singularities
Easy of gain tuning
Real-time
Multiple tasks
Rigorous stability analysis
Gianluca Antonelli Tokyo, 11.12.2014
Raising the abstraction
Gianluca Antonelli Tokyo, 11.12.2014
UW case study
a) End-effector configuration (dim 6)
b) Vehicle roll & pitch (dim 2)
c) Arm manipulability (dim 1)
Gianluca Antonelli Tokyo, 11.12.2014
UW case study
End-effector field of view (dim 2)
Vehicle roll & pitch (dim 2)
Arm manipulability (dim 1)
Mechanical joint limits (dim 7)
Gianluca Antonelli Tokyo, 11.12.2014
Aerial case study
Various actions embedding:
End-effector
Obstacle avoidance
Arm manipulability
Mechanical joint limits
Gianluca Antonelli Tokyo, 11.12.2014
ARCAS final objectives
Gianluca Antonelli Tokyo, 11.12.2014
ARCAS preliminary experiments
Gianluca Antonelli Tokyo, 11.12.2014
To do
Set-based (unilateral, inequality...) control
With Signe Moe & Kristin Pettersen, NTNU
Task-priority inverse kinematics vs dynamics
With Ciro NataleSUN
chattering
Humanoids?
Gianluca Antonelli Tokyo, 11.12.2014
AcknowledgeThanks to the colleagues from the
University of Cassino:Filippo Arrichiello, Khelifa Baizid, Elisabetta Cataldi, Stefano Chiaverini, Amal Meddahithe Consortium ISME and PRISMAthe projects ARCAS, AEROARMS,
DexROV and MARIS