Pedro Batista WP2 Single & multiple vehicle control 2nd Field Training Workshop in Underwater Robotics Intervention Marine Robot and Dexterous Manipulatin for Enabling Multipurpose Intevention Missions Instituto Superior Técnico Dynamical Systems and Ocean Robotics Laboratory / ISR / LARSyS
36
Embed
Marine Robot and Dexterous Manipulatin for Enabling Multipurpose ... · PDF fileDexterous Manipulatin for . Enabling Multipurpose . Intevention Missions . Instituto Superior Técnico
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Pedro Batista
WP2 Single & multiple vehicle control
2nd Field Training Workshop in Underwater Robotics Intervention
Marine Robot and Dexterous Manipulatin for Enabling Multipurpose Intevention Missions
Instituto Superior Técnico Dynamical Systems and Ocean Robotics Laboratory / ISR / LARSyS
Underactuated AUV moving in a scenario where there is a fixed transponder (the target)
WP2 Single & Multiple Vehicles Control
Homing - problem statement
Design the control law such that the vehicle is driven toward a well defined neighborhood of the target
WP2 Single & Multiple Vehicles Control
Homing – progress report (1)
Step 1 Theoretical framework
full under-actuated nonlinear dynamics of an AUV in 3-D.
set within a sensor-based framework considering an USBL acoustic positioning system
reported in a journal paper (Int. Journ. of Robust and Nonlinear Control)
WP2 Single & Multiple Vehicles Control
Homing – progress report (2)
Step 2 Adaptation to
Nessie
Control law adapted for Nessie’s autopilot with linear and angular velocity commands
Special care was taken tuning the gains and adapting the control law, in simulation, to avoid instability or undesired behaviors (oscillations,
control saturations, bang-bang situations, etc.)
Algorithms were simplified to 2-D (independent depth controller is provided by Nessie)
WP2 Single & Multiple Vehicles Control
Homing – progress report (3)
Step 3 Preliminary
Experimental validation
Experiments with Nessie during the “TRIDENT Fall School: 1st Field Training Workshop in Underwater Robotics Intervention”
Transponder position simulated in-the-loop
Integration in ROS of the control laws that were developed and tested in Matlab
WP2 Single & Multiple Vehicles Control
Homing – experimental results (1)
Nessie Homing @ Roses, 2011/10/19, 11:03:32
Run 1
Run 2
Run 3
WP2 Single & Multiple Vehicles Control
Homing – progress report (4)
Step 4 Experimental
validation
Repeat trials from Roses to check repeatability
Adopt in-house built inverted USBL
Development of Body-fixed filter for the transponder position based on USBL and DVL
Successful evaluation of the control performance
WP2 Single & Multiple Vehicles Control
Homing – experimental results (2)
Virtual transponder
Nessie Run 1 Homing @ Loch Earn, 2012/04/23, 18:34:04
Steady heading
WP2 Single & Multiple Vehicles Control
USBL integration on Nessie
In Lab
Transponder
USBL mounted on Nessie: - Reconfigured reception array to fit with Nessie’s DVL and thrusters
Fully integrated on Nessie: - Mechanically and buoyancy - Electrically - Network - Software (ROS)
WP2 Single & Multiple Vehicles Control
Homing – experimental results (3)
Inverted USBL
USBL Raw fixes with outlier detector
Filtered transponder position in Body frame
Nessie Run 3 Homing @ Loch Earn, 2012/04/26,12:30:26
WP2 Single & Multiple Vehicles Control
Homing – future work
Dissemination
Transition to the simulator and the G-500 should be straightforward as it has been previously
implemented in ROS with Nessie.
Submit a joint journal paper with HWU with the experimental results.
Re-run algorithms, in a larger mission scenario, from multiple initial and interesting conditions with the transponder installed on
Delfim
Docking between two transponders
Alternatively One transponder and direction vector via acoustic modem.
Scenario
Problem: Steer the vehicle toward the final position, along the path defined by the final desired direction.
WP2 Single & Multiple Vehicles Control
Docking – problem statement
WP2 Single & Multiple Vehicles Control
Docking – solution overview
Docking
When the vehicle is still far away, under-actuated dynamics are considered and
the vehicle is driven faster
When the vehicle is close to the base, fully actuated dynamics and model
uncertainty are considered
Convergence is always guaranteed
Two-step solution
WP2 Single & Multiple Vehicles Control
Docking – progress report (1)
Step 1 Theoretical framework
full under-actuated nonlinear dynamics of an AUV in 3-D.
set within a sensor-based framework considering an USBL acoustic positioning system
progress has already been presented at the 2012 ACC theoretical framework to be published in a journal paper
WP2 Single & Multiple Vehicles Control
Docking – simulation results
The vehicle approaches the base with the correct attitude for docking operations.
Initial stage at constant surge velocity
All velocities decrease as vehicle approaches the docking position.
WP2 Single & Multiple Vehicles Control
Docking – progress report (2)
Step 2 Adaptation to
Nessie
Control law adapted for Nessie’s autopilot with linear and angular velocity commands
Special care was taken tuning the gains and adapting the control law, in simulation, to avoid instability or undesired behaviors (oscillations,
control saturations, bang-bang situations, etc.)
Algorithms were simplified to 2-D (independent depth controller is provided by Nessie)
WP2 Single & Multiple Vehicles Control
Docking – progress report (3)
Step 3 Experimental
validation
Experiments with Nessie (joint work with HWU)
Transponder position simulated in-the-loop
Inverted USBL + body-fixed filter
Direction of arrival is provided virtually in both cases
Integration in ROS of the control laws that were developed and tested in Matlab
WP2 Single & Multiple Vehicles Control
Docking – experimental results (1)
Virtual transponder and direction of arrival
Nessie Run 4 Docking @ Loch Earn, 2012/04/24, 12:22:59
Final Docking direction: South (Yaw = 180 deg)
Docking controller
PID controller
WP2 Single & Multiple Vehicles Control
Docking – experimental results (2)
Inverted USBL, virtual direction of arrival
USBL Raw fixes with outlier detector
Filtered transponder position in Body frame Nessie Run 4 Docking @ Loch Earn, 2012/04/26,
15:14:08
Final Docking direction: South (Yaw = 180 deg)
WP2 Single & Multiple Vehicles Control
Docking – future work
Dissemination
Again, integration in the simulator and G-500 should be straightforward from ROS.
Submit a joint journal paper with Heriot-Watt with the docking experimental results.
Re-run algorithms, in a larger mission scenario, from multiple initial and interesting conditions with the transponder installed on
Delfim. The direction of arrival will be provided by Delfim.
WP2 Single & Multiple Vehicles Control
Bottom–Following – problem statement
Design a controller to achieve smooth tracking of the bottom profile, at a fixed distance, with feed-forward preview control to help limit the actuation bandwidth.
Problem Statement
General Idea
Path to be followed generated from sensor readings (depth sensor,
look-ahead sensor)
Path-following problem converted to regulation problem by careful definition of the error space
Design appropriate controller to drive tracking error to the origin
WP2 Single & Multiple Vehicles Control
Bottom–Following – summary
Collect sensor data to draw the bottom profile, add elevation offset to track it from a safe distance.
Sensor readings
WP2 Single & Multiple Vehicles Control
Bottom-Following – Path generation (1)
Bottom profile is approximated by straight line segments, which compose the path to be followed.
Path Generation
WP2 Single & Multiple Vehicles Control
Bottom-Following – Path generation (2)
Controller Design Process
Inclusion of the preview information in the error dynamics through state augmentation
Linearization and discretization of the error dynamics around several operating points
H2 controller design with gain scheduling to achieve good performance in all operating regions
D-methodology (integrators moved to the plant input) to achieve auto-trimming
WP2 Single & Multiple Vehicles Control
Bottom-Following – controller design (1)
Integration with the Nessie AUV, in cooperation with the group from
Heriot-Watt University
First Experimental Trials
A few adaptations were made in the controller that was implemented on Nessie
The vehicle was controlled only in surge and heave, to reduce stress on the vertical thrusters (pitch control was deprecated)
Preview was not included due to the resulting loss in directionality of the error space and the low velocities practiced in the trials
Five operating regions, depending on the slope of the path
WP2 Single & Multiple Vehicles Control
Bottom-Following – experimental setup
Generating altitude measurements from a predefined virtual path
First Trial
WP2 Single & Multiple Vehicles Control
Bottom-Following – Trial 1: virtual path
Going away from the shore, towards the middle of the loch
Second Trial
WP2 Single & Multiple Vehicles Control
Bottom-Following – Trial 2: real terrain
Coming back towards the shore, from the middle of the loch
Third Trial
WP2 Single & Multiple Vehicles Control
Bottom-Following – Trial 3: real terrain
Scenario
Theoretical frameworks and solutions have been developed
On a first approach, control for single vehicles was addressed
Theoretical solutions will continue to be adapted to practice in the near future…
Very encouraging preliminary experimental results on leader-following with acoustics and with inverted USBL
WP2 Single & Multiple Vehicles Control
Leader-Following – Preliminary results (1)
WP2 Single & Multiple Vehicles Control
Leader-Following – Preliminary results (2)
Nessie dead-reckoning and receiving ASC position.
Ad-Hoc (based on homing controller) leader-following with inverted USBL tracking the transponder on a moving boat (ASC) and without acoustic communications.
WP2 Single & Multiple Vehicles Control
In this workshop…
Experimental validation of all algorithms with Nessie and Delfim
Full USBL integration
Body-fixed navigation filter
Homing
Docking
Bottom-following
Leader following
Pedro Batista
WP1 Navigation and Mapping
2nd Field Training Workshop in Underwater Robotics Intervention
Marine Robot and Dexterous Manipulatin for Enabling Multipurpose Intevention Missions
Instituto Superior Técnico Dynamical Systems and Ocean Robotics Laboratory / ISR / LARSyS
Navigation For teams of vehicles
Single vehicle
Linear motion GAS
USBL, LBL
Reduced measurements
Range-based Bearings only
Range-only
Single and multiple
Bearings-only
Source localization and navigation
Relative position
USBL with biased DVL USBL with accelerometers
Absolute position
LBL with biased DVL LBL with accelerometers
Angular motion GES
Vector-based
Cooperative Magnetometer
unavailable
Angular motion GAS
USBL + Modem RG bias included
Linear motion GAS
Single-range USBL
Over 10 Journal articles: JIRS, TRO, EJC, IJC, AUTOMATICA (3), SCL (3), CEP, etc. Over 20 Conference publications: CDC, ACC, IFAC WC, ICRA, etc.
WP1 Navigation & Mapping
T1.1 Cooperative Navigation
Linear motion GAS
Relative position
USBL with biased DVL USBL with accelerometers
Absolute position
LBL with biased DVL LBL with accelerometers
Range-based
Single ranges - rich trajectories Multiple ranges
Bearings-only
USBL based Observability analysis
USBL Best with DVL!
LBL is complex to operate!
Rich trajectories required mostly. Possible easier
extension - depth measurements.
Also needs rich trajectories.
- LTI solution - Simple gain - GAS - Great performance