Show EOL distros: 

shadow_robot: sr_description | sr_example | sr_gazebo_plugins | sr_hand | sr_hardware_interface | sr_kinematics | sr_mechanism_controllers | sr_mechanism_model | sr_move_arm | sr_movements | sr_robot_msgs | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

  • Author: Ugo Cupcic / ugo@shadowrobot.com , contact@shadowrobot.com
  • License: GPL
  • Source: bzr lp:sr-ros-interface/stable (branch: release)
shadow_robot: sr_description | sr_example | sr_gazebo_plugins | sr_hand | sr_hardware_interface | sr_kinematics | sr_mechanism_controllers | sr_mechanism_model | sr_move_arm | sr_movements | sr_robot_msgs | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

  • Author: Ugo Cupcic / ugo@shadowrobot.com , contact@shadowrobot.com
  • License: GPL
  • Source: bzr lp:sr-ros-interface/fuerte (branch: release)
shadow_robot: sr_description | sr_example | sr_gazebo_plugins | sr_hand | sr_hardware_interface | sr_mechanism_controllers | sr_mechanism_model | sr_move_arm | sr_movements | sr_robot_msgs | sr_self_test | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

shadow_robot: sr_description | sr_example | sr_gazebo_plugins | sr_hand | sr_hardware_interface | sr_mechanism_controllers | sr_mechanism_model | sr_moveit_config | sr_movements | sr_robot_msgs | sr_self_test | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

shadow_robot: sr_description | sr_example | sr_hand | sr_hardware_interface | sr_mechanism_controllers | sr_mechanism_model | sr_movements | sr_robot_msgs | sr_self_test | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

sr_core: sr_gazebo_sim | sr_hand | sr_hardware_interface | sr_mechanism_controllers | sr_mechanism_model | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand:(fake) calibration controllers, position controllers, velocity controllers, force controllers, ...

sr_core: sr_gazebo_sim | sr_hand | sr_hardware_interface | sr_mechanism_controllers | sr_mechanism_model | sr_tactile_sensors | sr_utilities

Package Summary

The sr_mechanism_controllers package contains different types of controllers for the etherCAT hand: position controllers, velocity controllers, force controllers, ...

Overview

The different controllers are implemented in this package. The controllers have access to an actuator which contains the hand data updated in real time by sr_edc_ethercat_drivers and sr_robot_lib.

Each controller is sending force demands to a motor at 1kHz (an internal force control loop is running at 5kHz on the motor boards).

Friction Compensation

A friction compensation algorithm is available for each controller. The idea behind the friction algorithm is very simple: due to the tendon friction, the force to apply to move a joint is a function of the current joint position and the direction which you want to take.

The sr_friction_compensation package is used to record and interpolate the friction compensation map. Based on this map, we just add a value to the output of the controller as illustrated below (we have two different maps, one for each direction): friction_compensation.png

Hysteresis Deadband

In our controllers, we're using what we call an hysteresis deadbands for avoiding to drive the motors unnecessarily: when we're entering a small deadband, if the error changes sign, we're not driving the motor, unless we leave a bigger deadband: hysteresis.png

Controller Types

Effort Controllers

This is a very simple controller, as the real force control loop is done on the motor board. The force demand is directly transmitted to the motor, after limiting it to a range specified in the parameters and adding the friction compensation, except in the specified deadband.

Velocity Controllers

This is a PID controller transforming a velocity demand into a force demand. First (if we're outside of a given deadband) the velocity error is fed to a pid loop. The friction compensation is then added to the output of this loop (if we're outside of a certain deadband), and the motor demand is limited to the range specified by the user.

Position Controllers

This is a PID controller transforming a position demand into a force demand. First (if we're outside of a given deadband) the position error is fed to a pid loop. The friction compensation is then added to the output of this loop (if we're outside of a certain deadband), and the motor demand is limited to the range specified by the user.

Mixed Position / Velocity Controllers

This controller is a position controller on top of a velocity controller. This controller ensures that the joint travels at a given speed, and slows down to stop on the target.

Based on the position error, a velocity demand is generated from a first PID loop.

The velocity demand is then fed to the velocity controller described previously, which generates a force demand to send to the motor through a second PID loop.

Wiki: sr_mechanism_controllers (last edited 2011-11-07 12:03:43 by UgoCupcic)