[[TracNav]][[TOC]] = Introduction / はじめに = This rulebook addendum provides a competition in the !RoboCup logistics league domain, where the main tasks of the league are captured in isolated challenges.[[br]] このルールブックは，!RoboCup logistics leagueの競技であり，リーグの主なタスクが独立した課題に取り込まれた競技を提示します． The objectives of this new competition are:[[br]] この新しい競技の目的は次の通りです． * to provide a framework that allows teams to show and evaluate their progress in the individual tasks of the RCLL * to ease the preparation for the main competition through providing a simplified cost- and space-efficient setup suitable for replication in local labs * to be attractive for both !RoboCup live events and online competitions, where teams can participate remotely from all over the world * チームがRCLLの個々のタスクの進捗を表示および評価できるフレームワークを提供する * 各研究室での（競技の）複製に適した，コストとスペースが簡素化されたセットアップを提供することで，主要な競技の準備を容易にする * チームが世界中からリモートで参加できるRoboCupライブイベントとオンラインコンテストの両方に魅力的である This document is not self-sufficient, instead all the rules from the main [https://github.com/robocup-logistics/rcll-rulebook/releases rulebook] apply, unless stated otherwise. The proposed format is carried out at !RoboCup 2021, which is held remotely. Hence the remainder of this document establishes the rules around the online competition.[[br]] このドキュメントは独立したものではなく，特に明記されていない限り，メインの[https://github.com/robocup-logistics/rcll-rulebook/releases ルールブック]のすべてのルールが適用されます． 提案されたフォーマットは、リモートで開催されるRoboCup2021で実行されます。 したがって，このルールブックの残りの部分では，オンライン競技に関するルールを確立します． = Qualification / 参加資格 = In addition to the regular Qualification process, teams have to participate in a !RefBox [https://fh-aachen.sciebo.de/s/Qkm7VfIIEIIJk1d workshop] To complete the workshop, a game report dump of at least two Challenges has to be submitted to the OC.[[br]] 通常の選考過程に加えて，チームはRefBox[https://fh-aachen.sciebo.de/s/Qkm7VfIIEIIJk1d 勉強会]に参加する必要があります． ワークショップを完了するには，少なくとも2つのチャレンジのゲームレポートを組織委員会（OC）に提出する必要があります． [[br]]※ 今回のRCAP2021では，世界大会参加経験チームのみが参加しているので，この部分は省略しています． = Requirements to score points / 得点を獲得するための要件= For your challenges to count as completed and score points you need to submit[[br]] チャレンジを完了としてカウントし、ポイントを獲得するには、 1. the recorded video footage 1. your game reports from the MongoDB a. 記録されたビデオ映像 a. MongoDBによるゲームレポート to the OC.[[br]] をOCに提出する必要があります． Your submission will not be counted if the footage shows human interference with the field, machines or robots outside of the specified rules in Section 5.3.[[br]] 5.3節で指定されたルール以外で，人によるフィールド・機械・またはロボットへの干渉を映像が示している場合，提出物はカウントされません． = Competition Area / 競技エリア = == Field Layout / フィールドのレイアウト == [[Image(figure1.jpg, 320px, align=right)]] The competition area for the main challenges consists of a 5x5 [m] area divided in square zones of 1x1 [m]. Additional challenges that are not counting towards the scoring of the competition are carried out on a 7x8 [m] field instead (corresponding to half the field of the regular RCLL field). The field is a subset of the magenta-half of the regular field as depicted in Figure 1.[[br]] メイン課題の競技エリアは，5x5 [m] のエリアで1x1 [m] の正方形で分割されています． 競争の得点にカウントされない追加のチャレンジは，代わりに7x8 [m]のフィールドで実行します（通常のRoboCup Logistics League（RCLL）のフィールドの半分に対応します）． このフィールドは，図1に示すように、通常のフィールドのマゼンタ側の半分です． The entire area belongs to a single team. The bottom left-most 1x3 [m] area is called the insertion zone and provides the starting positions for up to three robots. Contrary to the full RCLL field, it is not necessary to build partial walls around the insertion zone and robots may be move freely in and out of the insertion zone.[[br]] エリア全体が1つのチームに割り当てられています． 左下の1x3 [m]の領域は挿入ゾーンと呼び，最大3台のロボットの開始位置になります． 通常のRCLLのフィールドとは異なり，挿入ゾーンの周囲に部分的な壁を構築する必要はなく，ロボットは挿入ゾーンの内外を自由に移動できます． Teams can play as either cyan or magenta, but the field is always the magenta one depicted in \reffig{fig:challenge-field}. Machine positions published by the refbox place the machines of both teams on the same location (e.g., M-BS and C-BS are both placed at M_Z12 in the field of Figure 2. == Mockup Machines == In case no real \ac{MPS} stations are available, replications (so called \emph{mockup machines}) may be used, that do not need to physically perform the respective production steps. Instead, that work may be carried out by a human supervisor (see \refsec{sec:operators}). The minimum requirements for a mockup machine are specified in the following. Mockup machines are required to have the same box-like base shape as specified in the RCLL rulebook. Additionally, the following parts need to be mounted: \begin{itemize} \item a model of the conveyor belt \item a shelf on the front right side of the box on stations replicating a \ac{CS} \item either a shelf, a slide or a conveyor belt on the front right side of the box, such that it is accessible from the front on stations replicating a \ac{RS} \end{itemize} Models for a conveyor belt, shelf and slides can be found in the RCLL rulebook repository\footnote{\url{https://github.com/robocup-logistics/rcll-rulebook/tree/common/challenges/mock_up_models}}. % chktex ignore-long-line The building materials for the models must be opaque, but may have any color. In order to compete in all main challenges, a minimum number of 3 mockup machines are required. On higher difficulty some challenges may require 4 mockup stations and some secondary challenges require 7 stations. == Remote Setup == In case a competition is carried out remotely, a proper local setup has to be established and approved by the \ac{OC}. Requirements include a proper camera setup that covers the field sufficiently, such that external viewers can verify the integrity of each challenges, as well as an approval for every mockup machine and robot that is used. After registration the \ac{OC} will verify the field setup in a video call. The validation call will be scheduled individually for each team to account for timezones. When challenges are played remotely, a video recording of the live-streams hast to be sent to the \ac{OC}. = Game Play = == Competition Scope == Each team can attempt a challenge by specifying a 30 minute time slot along with the challenge in advance of the respective competition day. The announcement procedure is managed by the \ac{OC}. Each slot can be used to do at most one challenge and is split in two phases: 1. In the \texttt{SETUP} phase the participants ensure that * the video stream is setup via Zoom, * the RefBox is started with the appropriate settings for the challenge, * the machines and robots are placed on the field according to the rules * and the robots are started. The setup must be completed within 10 minutes, otherwise the attempt is counted as a failure. 1. Then the \texttt{PRODUCTION} phase starts by instructing the RefBox accordingly. During the production phase the team has 20 minutes to solve the challenge. A team may retry the challenge by 1. setting the RefBox back to the setup phase 1. reesetting the field setup accordingly 1. and then starting the production phase again. Only the last attempt in each slot can be counted. A team may decide to count no attempt at all. Once a challenge is counted, it may not be retried again, unless the difficulty is increased. In order to complete an attempt, the RefBox needs to be set to phase \texttt{POST\_GAME}. All challenges are conducted while measuring the execution time of the corresponding attempt, starting from the begin of the \texttt{PRODUCTION} phase and ending at the begin of \texttt{POST\_GAME}. The fastest team to complete a challenge on the selected difficulty gains $5$ additional points. A challenge is only completed if the full score is achieved, so getting only partial credits for a challenge disqualifies from getting the bonus points for the fastest completion. == Changes compared to the Main Competition == The tasks covered in the various challenges have to be executed following the regular rules for the \ac{RCLL}, unless stated otherwise. However, some aspects are altered to simplify the setup. The changes are not affecting existing machine communication and processing steps, such that the challenges can be carried out on real machines as well as on mockup versions obeying the requirements outlined in \refsec{sec:machines}. === Product Delivery === The delivery procedure for finished products is altered compared to the \ac{RCLL} rule set. In order to reduce the amount of machines required for participation, Deliveries are made by bringing the finished product to the insertion zone and dropping it there. === Ring Payments === Easing the setup of mockup machines, it is not required to have a slide on ring stations. Instead, a shelf or additional conveyor may be mounted on the machine to place payments at the corresponding station. === Ring Color Assignment === The cost for mounting each ring color are fixed, the assignment of ring colors is semi-fixed as teams can choose between two different options for each challenge (\texttt{option1} or \texttt{option2} according to \reftab{tab:ring-costs}). The RefBox settings default to \texttt{option1}, but teams may change this accordingly. \newcommand{\colconfig}{\mathcal{RC}} \begin{table}[!htb] \centering \begin{tabular}{l|l|l||l|l||l|l} & \multicolumn{4}{c||}{Ring Costs} & \multicolumn{2}{c}{\multirow{2}{*}{Color Assignment }}\\\cline{2-5} & Color & Price & Color & Price & \multicolumn{2}{c}{}\\\cline{2-7} & Yellow & 0 & Green & 0 & RS1: $\colconfig_1$ & RS1: $\colconfig_2$ \\ & Blue & 1 & Orange & 2 & RS2: $\colconfig_2$ & RS2: $\colconfig_1$ \\\hline\hline Configuration & \multicolumn{2}{c||}{$\colconfig_1$} & \multicolumn{2}{c||}{$\colconfig_2$} & $\texttt{option1}$ & $\texttt{option2}$\\ \end{tabular} \caption{Materials} \label{tab:ring-costs} \end{table} === Materials === The available material that can be used per challenge is restricted (unless stated otherwise) per machine according to the information in \reftab{tab:materials}. \begin{table}[!htb] \centering \begin{tabularx}{\linewidth}{l|l} Machine & Available Material \\\hline \ac{BS} & 2 bases of each color \\ \ac{CS} & 3 cap-carriers (cap color choices up to each team) \\ \ac{RS} & 4 rings of each assigned color (8 in total) \\ \end{tabularx} \caption{Materials} \label{tab:materials} \end{table} === Orders === Unless specified otherwise, orders that have to be fulfilled in challenges are entered through the web shop (see the RefBox workshop in \refsec{sec:qualification}) by any member of the team. In challenges where only one \ac{RS} is present, teams are responsible to order products which can be assembled using the available stations only. === Scoring === While the RefBox may assign points during challenges according to the regular RCLL rules, those points do not count towards this competition. Instead, the point scoring for each challenge is listed in \refsec{sec:challenges}. == Field Operators == In challenges where mockup machines are used, the actual assembly steps have to be performed by human supervisors. Whenever a machine is instructed, the RefBox operator announces the required interaction. One field operator may proceed to enter the field in order to perform the interaction. Afterwards the field has to be left immediately. The usual rules for replenishing resources (respecting the limited materials \refsec{sec:materials}) apply. == Available Challenges for the Primary Competition == Challenges have different types and variations (difficulty levels). The overall score of the competition is calculated by summing up the score in the highest difficulty achieved in each of the challenge types. The challenge types of the competition are described in \refsec{sec:challenge-navigation}-\ref{sec:challenge-markerless}. The RefBox is used to log the progress and data for each challenge. Participating teams are required to send the corresponding game report dumps to the \ac{OC}. Once the competition is finished, $5$ bonus points are awarded each time a team solved a challenge on a difficulty in the shortest amount of time. == Navigation Challenge == Basic navigation task with known obstacles.\\ \textbf{Task:} Drive to 12 randomly generated target zones. A target zone is reached, if any robot remains in that zone for at least 5 consecutive seconds. The pose of each robot must be set in the beacon message for the RefBox to properly register the zones as visited. The robot may move within the zone during the 5 second period.\\ Variations of this challenge depend on the number of available machines (see \reftab{tab:challenge-navigation}). Multiple robots may be used to simultaneously to reach multiple target zones. Partial points may be awarded in case only a subset of target zones were reached, which are accumulated if the challenge is completed fully. \begin{table}[!htb] \centering \begin{tabular}{l|l|l|l|l} \multirow{2}{*}{Machines} & \multicolumn{4}{c}{Scoring} \\\cline{2-5} & $\geq$ 4 zones & $\geq$ 8 zones & all 12 zones & combined \\\hline\hline 2 & +2 & +4 & +9 & 15 \\ 3 & +3 & +6 & +14 & 23 \\ 4 & +4 & +8 & +18 & 30 \\ \end{tabular} \caption{Navigation Challenge} \label{tab:challenge-navigation} \end{table} == Exploration Challenge}\label{sec:challenge-exploration == Replicate the RCLL exploration phase. Machine Marker detection as well as navigational skills are required to solve this challenge.\\ \textbf{Task:} Find and report all machines on the field (type and orientation) according to the rules of a regular exploration phase. \\ Variable in the number of machines (see \reftab{tab:challenge-exploration}). \begin{table}[!htb] \centering \begin{tabular}{l|l} Machines & Scoring \\\hline 2 & 10 \\ 3 & 20 \\ 4 & 30 \\ \end{tabular} \caption{Exploration Challenge} \label{tab:challenge-exploration} \end{table} This challenge needs to be run by setting the RefBox to phase \texttt{EXPLORATION}. === Grasping Challenge}\label{sec:challenge-grasping === Simple grasping task. Each Machine has a base at output. Robots start at the zone in front of a machine output.\\ \textbf{Task:} A robot brings a base from one machine's output back to its input. A human supervisor places it back to the output. Repeat until all products were placed at the respective machines input 3 times and all robots returned to their starting positions. \\ Variations differ by number of machines, see \reftab{tab:challenge-grasping}. The $i$-th repetition is considered to be successful, once all bases were placed at the respective machine input at least $i$ times. The placement of the machines on the field is fixed according to \reffig{fig:grasping-field}. In order to load the fixed field shown in \reffig{fig:grasping-field}, a corresponding game report\footnote{\url{https://github.com/robocup-logistics/rcll-refbox/blob/tviehmann/challenges-2021/benchmarks/grasping_challenge.gz}} % chktex ignore-long-line can be restored in the RefBox using the supplied script.\footnote{\url{https://github.com/robocup-logistics/rcll-refbox/blob/tviehmann/challenges-2021/etc/scripts/restore_reports.bash}} % chktex ignore-long-line In order to load the field setup, specify tha report to laod from and disable field randomization in the RefBox configuration as shown in \reflst{lst:restore-field}. \begin{lstlisting}[language=yaml,basicstyle=\small, caption={RefBox configuration to load grasping challenge field}, captionpos=b, label={lst:restore-field}, frame=single] random-field: false ... # load data from latest game report with a given name # leave empty to always load from latest stored report load-from-report: "GraspingChallenge" \end{lstlisting} \begin{table}[!htb] \centering \begin{tabular}{l|l|l} \multirow{2}{*}{Machines} & \multicolumn{2}{c}{Scoring} \\\cline{2-3} & first repetition & each subsequent repetiton \\\hline\hline 1 & +10 & +2 \\ 2 & +20 & +2 \\ 3 & +25 & +2 \\ \end{tabular} \caption{Grasping Challenge} \label{tab:challenge-grasping} \end{table} \begin{figure} \centering \begin{tikzpicture}[x=-1cm,y=1cm] \fill[gray!50!white] (2,0) rectangle (5,1); \node at (3.5,0) [anchor=south, font=\tiny, inner sep=1pt] {Insertion Zone}; \foreach \i in {0,...,5} { \draw (0,\i) -- (5,\i); \draw (\i,0) -- (\i,5); } \foreach \x in {1,...,5} { \foreach \y in {1,...,5} { \node[font=\tiny,anchor=north west, inner sep=1pt] at (\x,\y) {M\_Z\x\y}; } } \node at (4.5,1.5) [circle,fill=magenta, draw=black,inner sep=0.2cm] {}; \node at (4.5,2.5) [rectangle,fill=green!30!white, draw=black, inner sep=0.1pt,align=center,font=\tiny] {input\\output}; \node at (0.5,0.5) [circle,fill=magenta, draw=black,inner sep=0.2cm] {}; \node at (0.5,1.5) [rectangle,fill=green!30!white, draw=black, inner sep=0.1pt,align=center,font=\tiny] {input\\output}; \node at (0.5,4.5) [circle,fill=magenta, draw=black,inner sep=0.2cm] {}; \node at (1.5,4.5) [rectangle,fill=green!30!white, draw=black, inner sep=0.1pt,align=center,font=\tiny,rotate=90] {input\\output}; \end{tikzpicture} \caption{Starting configuration for the grasping challenge. BS is placed at $\texttt{M\_Z12}$, CS1 is placed at $\texttt{M\_Z53}$ and RS1 is placed at $\texttt{M\_Z25}$.} \label{fig:grasping-field} \end{figure} === Product Challenges === This section covers four types of challenges, which all can be individually completed. Each challenge corresponds to the production of a product with one of the available complexities (C0, C1, C2, C3) in the \ac{RCLL} using either one or two \ac{RS}.\\ For complexities C1, C2 and C3 the accumulated cost for mounting the required rings must be equal to 1, 2 and 3, respectively. \textbf{Task:} Produce all posted orders.\\ \begin{table}[!htb] \centering \begin{tabular}{l|l|l} Machines & Challenge type & Scoring \\\hline 2 & C0 & 30 \\ 3 & C1 & 50 \\ 4 & C1 & 50 \\ 3 & C2 & 70 \\ 4 & C2 & 70 \\ 3 & C3 & 100 \\ 4 & C3 & 100 \\ \end{tabular} \caption{CX Challenge} \label{tab:challenge-cx} \end{table} === Exploration + Production Challenges === The same challenge as in \refsec{sec:challenge-cx} but without receiving the machine positions. Hence exploration and production takes place concurrently. The same point scores as in \reffig{tab:challenge-cx} apply with an additional $+20$ per challenge === Markerless Detection Challenge === Image recognition challenge to classify different machine types.\\ \textbf{Task:} Autonomously label the machines shown in a set of pictures by their type. A picture may contain more than one machine, which all count towards the total number of machines to detect. Labels must be placed on the machines within the picture. Machines are distinguished by their types (BS, CS, RS, DS, SS).\\ As a preparation for this challenge, a data set will be supplied to all participants which may be used for training and testing purposes\footnote{\url{https://fh-aachen.sciebo.de/s/MRghypMQSMhTYaC}}. The evaluation set for the challenge consists of a set of separate images that is encrypted and distributed in advance of attempting the challenge. The password to decrypt the evaluation set is given out when the challenge is attempted. This challenge can only be attempted once to avoid improper use of the evaluation set. The points scored are calculated based on the relative amount of correctly and wrongly classified machines (across all types) according to \reftab{tab:challenge-markerless}. So a total amount of 30 points can be scored if all machines of all types are correctly classified. \begin{table}[!htb] \centering \begin{tabularx}{\linewidth}{l|l|l|l} \% Correctly Classified & \% Wrongly Classified & \% Not Classified & Scoring \\\hline $x$ & $y$ & $z$ & $(x-y)\cdot30$ \end{tabularx} \caption{Machine Detection Challenge} \label{tab:challenge-markerless} \end{table} In order to verify the labeling, each team must presents the labeled data within the time slot (e.g., via screen-sharing when participating remotely). Additionally, the labeled data has to be sent to the \ac{OC}. == Challenges for the Secondary Competition == === Full game === Play the production phase of an RCLL game on a field of \SI{7 x 8}{\metre} with $7$ machines (no machines from the opposing team). The normal RCLL rules apply with the exception that the ground truth for machine positions is not sent by the RefBox.