[Spinal][PwmTest] workaround to send pwm value to each servo independependently

[sugihara-16]

What is this

Change some structure of the PWM test to support motor-by-motor check

Usage

• Send PWM value (0.55) to all motors simultaneously
  • rostopic pub -1 /pwm_test spinal/PwmTest "motor_index: [] pwms: [0.55]"
• Send PWM value (motor 0 → 0.55, motor 2 → 0.65) independently
  • rostopic pub -1 /pwm_test spinal/PwmTest "motor_index: [0,2] pwms: [0.55, 0.65]"
• Escape from PWM test mode
  • rostopic pub -1 /pwm_test spinal/PwmTest "motor_index: [] pwms: []"

[sugihara-16]

@tongtybj This PR doesn't depend on any other features, so could you please check and merge this?

[tongtybj]

LGTM.

But I do think it would be better to refactor Pwms.msg rather than defining a new message. On the other hand, I also understand this will effect the whole system if we change the structure of Pwms.msg.
Therefore, current solution is still temporary, right?

[sugihara-16]

In my understanding, Pwms.msg is for state message rather than for command, and I guess this is the reason why the current system doesn't use Pwms.msg for PWM test, right?
If so, I think the best solution is to provide status messages and command messages independently as well as the case of servo control. How do you think about it?

[tongtybj]

provide status messages and command messages independently

I was told that creating original message type is a stupid way, and using the standard message type such as std_msgs, geometry_msgs is the clever way. Considering the limited bandwidth of communication between PC and MCU, we have to define some original light-weight message types. But the number of original message type should be kept to the minimum. Therefore, if two different topics have very similoar contents, it is better to use a common message type.

I guess this is the reason why the current system doesn't use Pwms.msg for PWM test, right?

It is just becuase I didn't realize the need of sending different PWM values to different motor...

[sugihara-16]

OK, so the best way you think is using Pwms.msg for both status and command topics, right?

[tongtybj]

That only applies to half.

The best solution I suggest is to use PwmTest.msg (of couse, rename it to Pwm.msg or something like that ) for both status and command topics.

[sugihara-16]

All right, I'll modify some parts as you have suggested.

[tongtybj]

I'm sorry I didn't explain it enough...
I just want to share a consensus with you about the design message type.
Just I said, refactor of Pwms.msg would induce a large change to our system.
We can consider it as a future work when we decide to refactor the whole system.

I will merge this PR with current version.

[sugihara-16]

All right, everything is clear.
Thank you for your suggestions!