Recovery behavior

[rokusottervanger]

If an action failed, this does not mean the following actions are useless to do. However, this means that the action scheduling needs to be aware of dependencies between actions.

[LoyVanBeek]

A start would be to specify pre/post-conditions on actions.

[rokusottervanger]

That would indeed be a good start, but it would also include a (partial) state of the world. What would be a good way to represent that?

…

On May 20, 2017 14:40, "Loy" ***@***.***> wrote: A start would be to specify pre/post-conditions on actions. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#16 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AIlLB9x0icNOHGNz_hW6Jywet2zWdZleks5r7t8-gaJpZM4NhSUo> .

[LoyVanBeek]

A pre/post condition could be to require a certains tate of the world model, e.g. robot is in some place or knows that some entity is at some place.

Perhaps ROSPlan is interesting for this.