diff --git a/c2-scheduling-problem.tex b/c2-scheduling-problem.tex index 121caa0e033668b1b37fdf6ce84c3c2b7f05977c..0cbeca4027637bda6beaab01c17b09af07660163 100644 --- a/c2-scheduling-problem.tex +++ b/c2-scheduling-problem.tex @@ -4,7 +4,7 @@ With all the previous considerations in mind, we now define the C2 scheduling problem. In addition to what was said before, we make the following assumptions. No preemption: -Tasks can not be interrupted once they are started. +Tasks cannot be interrupted once they are started. Task durations are integer: Durations are unitless numbers which can stand for days, hours, or minutes. Fractions are not necessary. diff --git a/introduction.tex b/introduction.tex index 349f0b70ab11b1d5c7bd09b21d328226ad17f97a..8dba272dd80dc3a0d80360c38c74c49a5f7edb89 100644 --- a/introduction.tex +++ b/introduction.tex @@ -22,7 +22,7 @@ The long-term goal of this line of research is to develop methods for decision m Staying ahead of hectic situations requires two sorts of skills. On the one hand, it requires flexible and creative out-of-the-box thinking, something that humans are good at. On the other hand, it requires keeping track of large amounts of minute details, and quick assessment of different courses of actions, something that computers are good at. -Current state-of-the-art in the Dutch navy is that humans do everything by hand, with rudimentary or no computer support. +Current state-of-the-art in the Dutch navy is that humans do most of the task coordination by hand, with rudimentary or no computer support. The potential for improvement, even with simple tools, is immense. Decision making support provides tools for human-machine collaboration where each party can contribute their respective strengths. diff --git a/scheduling.tex b/scheduling.tex index a812b0e49320658f26e3767450088aa8194001c1..578226d6217c686e0d9bfb699aadcc057357db06 100644 --- a/scheduling.tex +++ b/scheduling.tex @@ -21,7 +21,7 @@ Such a chain of actions is called a \emph{plan}. Planning is only concerned with the effects of actions on the state, but not with who should execute them and when. This is the purpose of the scheduler. The output of the planner is the input of the scheduler. -In other words, planning answers the question What to do?'', while scheduling answers the question Who does it?'' +In other words, planning answers the question What to do?'', while scheduling answers the question Who will do it?'' There are two problems with this division. \paragraph{First} @@ -57,12 +57,12 @@ See \cref{sec:example-search-and-rescue} for an example. The scheduling problem that arises in C2 is a variant of the multi-skill resource-constrained project scheduling problem (MSRCPSP). This section describes how they relate. -The MSRCPSP is a variant of the RCPSP, which is itself a variant of the PSP. +The MSRCPSP \cite{Artigues2008} is a variant of the RCPSP, which is itself a variant of the PSP. In the basic project scheduling problem (PSP), each task requires specific resources to be executed. For example, the task \emph{review customer complaint} specifically requires Alice and Bob. The goal is to find a time slot where both Alice and Bob are available. -The resource-constrained PSP (RCPSP) generalizes the PSP in that each task can require a quantity of one or more of a class of resources \cite{Artigues2008}. +The resource-constrained PSP (RCPSP) generalizes the PSP in that each task can require a quantity of one or more of a class of resources. Every resource can only belong to exactly one class. For every class there is a pool of available resources. For example, it is possible to express that the task \emph{review customer complaint} requires one PR manager and two engineers.