Add new figure on skillrl for section6

section6/Figures/skillrl_illustration.tex

+% !Mode:: "TeX:UTF-8"
+% !TEX encoding = UTF-8 Unicode
+
+\begin{center}
+  \begin{tikzpicture}[remember picture]
+    \tikzset {
+      block/.style={draw,inner sep=0pt,fill=white, minimum width=5cm, minimum height=1.5cm, align=center},
+      miniblock/.style={draw,inner sep=0pt,fill=white, minimum width=1.5cm, minimum height=.8cm, align=center, rounded corners=2pt, text width=1.4cm},
+      linetext/.style={fill=#1, minimum height=1.5pt, minimum width=.8cm, inner sep=0},
+    }
+    \def\sep{1cm}
+    \def\ssep{1cm}
+    \begin{scope}
+      \node [block, anchor=north] (b1) at (0,0) {\footnotesize{Experience Collecting}};
+      \node [block, anchor=north] (b2) at ([yshift=-\sep]b1.south) {\footnotesize{Skill Generation}};
+      \node [block, anchor=north] (b3) at ([yshift=-\sep]b2.south) {\footnotesize{Skill Bank}};
+      \node [block, anchor=north] (b4) at ([yshift=-\sep]b3.south) {\footnotesize{Skill-Augmented RL Training}};
+      \node [block, anchor=north] (b5) at ([yshift=-\sep]b4.south) {\footnotesize{Recursive Skill Evolution}};
+
+      % \node [anchor=south] (b6) at ([yshift=\sep]b5.north) {\LARGE{$\cdots$}};
+      \draw [->] (b1.south) -- (b2.north);
+      \draw [->] (b2.south) -- (b3.north);
+      \draw [->] (b3.south) -- (b4.north);
+      \draw [->] (b4.south) -- (b5.north);
+      % \draw [->] (b5.north) -- (b6.south);
+
+      \scriptsize
+      \node [anchor=north west, text width=7.6cm] (n1) at ([xshift=\ssep]b1.north east) {Collect interaction trajectories from the enrionment.\\};
+      \node [anchor=north west, text width=3.5cm,draw,rounded corners,minimum height=1.5cm] (n12) at ([yshift=-0.1cm]n1.south west) {} ;
+      \node [anchor=west, text width=3.5cm,draw,rounded corners,minimum height=1.5cm] (n13) at ([xshift=0.3cm]n12.east) {} ;
+
+      \draw [dashed] ([xshift=0.8cm,yshift=0.1cm]b2.north east) -- ([xshift=9cm,yshift=0.1cm]b2.north east);
+      \node [anchor=north west, text width=7.6cm] (n2) at ([xshift=\ssep]b2.north east) {Generate reusable skills from experiences.\\};
+      \node [anchor=north west, text width=] (n21) at (n2.south west) {Successful trajectories};
+      \node [anchor=west, text width=] (n22) at ([xshift=.2cm]n21.east) {$\rightarrow$~~Skill Principles};
+      \node [anchor=north west, text width=] (n23) at (n21.south west) {Failed trajectories};
+      \node [anchor=north west, text width=] (n24) at (n22.south west) {$\rightarrow$~~Failed Lessons};
+
+      \draw [dashed] ([xshift=0.8cm,yshift=0.1cm]b3.north east) -- ([xshift=9cm,yshift=0.1cm]b3.north east);
+      \node [anchor=north west, text width=7.6cm] (n3) at ([xshift=\ssep]b3.north east) {Construct a hierarchical skill library.};
+      \node [anchor=north west, text width=7.6cm,draw,rounded corners,minimum height=0.8cm,align=left] (n31) at (n3.south west) {};
+      \node [anchor=west,text width=3cm,align=center] (n311) at ([xshift=0.2cm]n31.west) {\textbf{\small{General Skills}}};
+      \node [anchor=west,text width=4cm,align=left] (n312) at ([xshift=0.2cm]n311.east) {bla bla bla bla bla bla bla bla bla bla bla bla bla bla  };
+      \node [anchor=north west, text width=7.6cm,draw,rounded corners,minimum height=0.8cm,align=left] (n32) at ([yshift=-0.1cm]n31.south west) {};
+      \node [anchor=west,text width=3cm,align=center] (n321) at ([xshift=0.2cm]n32.west) {\textbf{\small{Task-specific Skils}}};
+      \node [anchor=west,text width=4cm,align=left] (n322) at ([xshift=0.2cm]n321.east) {bla bla bla bla bla bla bla bla bla bla bla bla bla bla  };
+      \draw [decorate,decoration={brace,mirror,raise=5pt,amplitude=6pt}] (n31.north west) -- (n32.south west);
+
+      \draw [dashed] ([xshift=0.8cm,yshift=0.1cm]b4.north east) -- ([xshift=9cm,yshift=0.1cm]b4.north east);
+      \node [anchor=north west, text width=7.6cm,align=left] (n4) at ([xshift=\ssep]b4.north east) {Retrieve relevant skills then provide retrieved skills to the agent.
+        Optimize the skill-augmented policy with RL.};
+      \draw ([yshift=-0.1cm]n4.south west) node (n41) [anchor=north west, text width=1.2cm,draw,rounded corners,minimum height=0.8cm,align=center] {Task}
+      ([xshift=0.8cm]n41.east) node (n42) [anchor=west, text width=1.2cm,draw,rounded corners,minimum height=0.8cm,align=center] {Retrieve \\ Skills}
+      ([xshift=0.8cm]n42.east) node (n43) [anchor=west, text width=1.2cm,draw,rounded corners,minimum height=0.8cm,align=center] {Agent}
+      ([xshift=0.8cm]n43.east) node (n44) [anchor=west, text width=1.2cm,draw,rounded corners,minimum height=0.8cm,align=center] {Reward};
+      \draw[->] (n41.east) -- (n42.west);
+      \draw[->] (n42.east) -- (n43.west);
+      \draw[->] (n43.east) -- (n44.west);
+
+      \draw [dashed] ([xshift=0.8cm,yshift=0.1cm]b5.north east) -- ([xshift=9cm,yshift=0.1cm]b5.north east);
+      \node [anchor=north west, text width=7.6cm,align=left] (n5) at ([xshift=\ssep]b5.north east) {Analyze failure trajectories during training.
+        Generate new skills or refine existing skills.};
+      \node [anchor=north west, text width=6cm,draw,dashed,rounded corners,minimum height=1cm,align=left] (n51) at (n5.south west) {};
+      \node [anchor=west,text width=2cm,align=center] (n52) at ([xshift=0.2cm]n51.west) {\textbf{\small{Outcome}}};
+      \path (n52.east) node(n53) [anchor=west, text width=1.3cm,draw,rounded corners,minimum height=0.6cm,align=center] {new skill}
+      ([xshift=0.5cm]n53.east) node(n54) [anchor=west, text width=1.3cm,draw,rounded corners,minimum height=0.6cm,align=center] {old skill};
+      \draw ([xshift=0.1cm]n53.south east) -- ([xshift=-0.1cm]n54.north west);
+    \end{scope}
+  \end{tikzpicture}
+\end{center}

section6/section6.tex

@@ -455,7 +455,8 @@ The above way of generating skills often suffer from quality issues. First, a ge
     % \resizebox{\linewidth}{!}{
     %     \input{section6/Figures/skillrl.tex}
     % }
-    \includegraphics[width=0.5\textwidth]{section6/Figures/skillrl_draft.png}
+    %   \includegraphics[width=0.5\textwidth]{section6/Figures/skillrl_draft.png}
+    \input{section6/Figures/skillrl_illustration.tex}
     \caption{
         Illustration of SkillRL \citep{xia-etal:skillrl}. SkillRL maintains a pool of agentic trajectories and initially abstracts reusable skills from these trajectories. The extracted skills are then used to guide RL training. As the agent improves through RL training, it generates higher-quality trajectories, which are further used to refine the skill bank.
     }