[FEA] Allow deactivating nodes when system is activated.

[ZhenshengLee]

There is a need to selectively disable the function of a node, that's say deactivating a node in a certain system mode.

Should the following config work? Thanks!

SHM file

---
# system modes example
---

actuation:
  ros__parameters:
    type: system
    parts:
      drive_base
      left/manipulator
      right/manipulator
    modes:
      __DEFAULT__:
        drive_base: active.__DEFAULT__
        left/manipulator: inactive.__DEFAULT__
        right/manipulator: inactive.__DEFAULT__
      LEFT:
        # drive_base: active.__DEFAULT__
        left/manipulator: active.__DEFAULT__
        right/manipulsator: inactive.__DEFAULT__
      RIGHT:
        # drive_base: active.__DEFAULT__
        left/manipulator: inactive.__DEFAULT__
        right/manipulator: active.__DEFAULT__


left/manipulator:
  ros__parameters:
    type: node
    modes:
      __DEFAULT__:
        ros__parameters:
          max_torque: 0.1

drive_base:
  ros__parameters:
    type: node
    modes:
      __DEFAULT__:
        ros__parameters:
          max_speed: 0.1
          controller: PID

right/manipulator:
  ros__parameters:
    type: node
    modes:
      __DEFAULT__:
        ros__parameters:
          max_torque: 0.11

step to reproduce

zs@zs-vm-2204:~/zs_ws/ga_ros$ ros2 service call /actuation/change_state lifecycle_msgs/ChangeState "{transition: {id: 1, label: configure}}"
waiting for service to become available...
requester: making request: lifecycle_msgs.srv.ChangeState_Request(transition=lifecycle_msgs.msg.Transition(id=1, label='configure'))

response:
lifecycle_msgs.srv.ChangeState_Response(success=True)

after this ,all nodes become OK

zs@zs-vm-2204:~/zs_ws/ga_ros$ ros2 service call /actuation/change_state lifecycle_msgs/ChangeState "{transition: {id: 3, label: activate}}"
requester: making request: lifecycle_msgs.srv.ChangeState_Request(transition=lifecycle_msgs.msg.Transition(id=3, label='activate'))

response:
lifecycle_msgs.srv.ChangeState_Response(success=True)

after this , every thing OK

zs@zs-vm-2204:~/zs_ws/ga_ros$ time ros2 service call /actuation/change_mode system_modes_msgs/ChangeMode "{mode_name: 'LEFT'}"
waiting for service to become available...
requester: making request: system_modes_msgs.srv.ChangeMode_Request(mode_name='LEFT')

after this, the mode_manager died with a segfault.

EDIT: @ralph-lange

the step of reproduce is updated in 2022-11-12.

[ralph-lange]

At first glance, this should work. You make even simply write drive_base: active instead of drive_base: active.__DEFAULT__ (as in right/manipulator: active). Did you experience any problems with this configuration?

[ZhenshengLee]

@ralph-lange thank you for your fast reply!

At first glance, this should work

In currently design, In any specific MODE(MODERATE and PERFORMANCE), all node must be in ACTIVE state, if any of the nodes become INACTIVE, the mode_inference will think the MODE is in DEACTIVATING state

Did you experience any problems with this configuration?

• The segfault happens with mode_manager with this config-yaml

I will repeat the test and give more details.