Updates to :stop-motion

[Affonso-Gui]

The :stop-motion stops the current angle vector by sending the actual angle vector as a new command (and therefore overwriting the old command).

The problem is that is also does that when there is no ongoing action, and since it also have an :wait-interpolation it takes a few seconds to return from the call.

In this PR we:

1. Only overwrite commands when there are ongoing actions
2. Add a wait key argument to the :stop-motion, so the user can opt not to wait-interpolation.

@pazeshun

[k-okada]

Do you need this because I did not merged start-jsk/rtmros_common#765 (comment) ?

[Affonso-Gui]

@k-okada This is not directly related to start-jsk/rtmros_common#765 (comment).

I noticed this problem when adding a call of :stop-motion to the default sigint behaviour, which led me to waiting a couple seconds every time C-c is pressed, even though there was no ongoing motion.

[YoheiKakiuchi]

Why this it also have an :wait-interpolation it takes a few seconds to return from the call. happens.

I saw the code, by this (send self :angle-vector av stop-time),
the duration of this command was limited by min-time (= 1.0 second).
https://github.com/jsk-ros-pkg/jsk_pr2eus/blob/master/pr2eus/robot-interface.l#L374

[k-okada]

so why not using :cancel-angle-vector -- ◉ Kei Okada 2019年10月8日(火) 14:48 Affonso, Guilherme <notifications@github.com>:

…

@k-okada <https://github.com/k-okada> This is not directly related to start-jsk/rtmros_common#765 (comment) <start-jsk/rtmros_common#765 (comment)> . I noticed this problem when adding a call of :stop-motion to the default sigint behaviour, which led me to waiting a couple seconds every time C-c is pressed, even though there was no ongoing motion. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#402?email_source=notifications&email_token=AADYNXFZIFX2GLCA452WSYDQNQNJNA5CNFSM4I6CVMH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEAS3P3Q#issuecomment-539342830>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADYNXCEKASYZ23CV5SK77TQNQNJNANCNFSM4I6CVMHQ> .

[Affonso-Gui]

@YoheiKakiuchi Sorry, saying that it takes a few seconds was a bit exaggerated. It seems to take exactly 1 second

$ (bench (send *ri* :stop-motion))
[ INFO] [1570526342.652453956]: wait-interpolation debug: start
[ INFO] [1570526343.678601001]: wait-interpolation debug: end
;; time -> 1.08321[s]
(nil nil nil nil)

@k-okada :cancel-angle-vector leads to an abrupt stop, while :stop-motion makes the robot stops smoothly. I thought that it would be best fitted for a demo to use the :stop-motion in planned interruptions.

[k-okada]

i see, please add that to doc string. -- ◉ Kei Okada 2019年10月8日(火) 18:25 Affonso, Guilherme <notifications@github.com>:

…

@YoheiKakiuchi <https://github.com/YoheiKakiuchi> Sorry, saying that it takes a few seconds was a bit exaggerated. It seems to take exactly 1 second $ (bench (send *ri* :stop-motion)) [ INFO] [1570526342.652453956]: wait-interpolation debug: start [ INFO] [1570526343.678601001]: wait-interpolation debug: end;; time -> 1.08321[s] (nil nil nil nil) @k-okada <https://github.com/k-okada> :cancel-angle-vector leads to an abrupt stop, while :stop-motion makes the robot stops smoothly. I thought that it would be best fitted for a demo to use the :stop-motion in planned interruptions. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#402?email_source=notifications&email_token=AADYNXHGB6STLYHGAFD2EHDQNRGY5A5CNFSM4I6CVMH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEATQ4KI#issuecomment-539430441>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADYNXHKVOHL5K3JDVPNKGDQNRGY5ANCNFSM4I6CVMHQ> .

[Affonso-Gui]

@k-okada Can I ask for a review?

[k-okada]

@Affonso-Gui how about adding “, uses are expected to use :cancel-angle-vector, for most cases” or something.

[Affonso-Gui]

Are uses expected to use :cancel-angle-vector for most cases?

[k-okada]

if it is supported, what is the reason to use :stop-motion ?

[Affonso-Gui]

Both :cancel-angle-vector and :stop-motion are (should be) supported on all robots that use follow-joint-trajectory controllers.

:cancel-angle-vector leads to an abrupt stop, while :stop-motion makes the robot stops smoothly.

Although there are robots in which :cancel-angle-vector is smooth enough, prioritizing speed or smoothness is task/motion dependent, so I don't believe we can generalize into using one over the other.

[k-okada]

I see, then how about adding "note that :cancel-angle-vector leads to an abrupt stop, while :stop-motion makes the robot stops smoothly.", ":stop-time is transition time for stopping motion, but the actual interpolation duration is limited by :min-time of :angle-vector, which is set to 1.0 [sec] by default" etc, to avoid confusion like my case.

[Affonso-Gui]

@k-okada Added more information to the docstring. Please check to see if it's clear enough.