placeholder for smoothing

Author: RussTedrake

book/index.html

@@ -484,6 +484,9 @@ <h1>Table of Contents</h1>
   <ul>
     <li><a href=planning.html#section1>Large-scale Incremental Search</a></li>
     <li><a href=planning.html#section2>Probabilistic RoadMaps (PRMs)</a></li>
+    <ul>
+      <li>Getting smooth trajectories</li>
+    </ul>
     <li><a href=planning.html#section3>Rapidly-exploring Random Trees (RRTs)</a></li>
     <ul>
       <li>RRTs for robots with dynamics</li>

book/planning.html

@@ -267,6 +267,7 @@ <h1><a href="index.html" style="text-decoration:none;">Underactuated Robotics</a
           &emsp;if {$\bq, \bq_n$} is collision free:<br/>
           &emsp;&emsp;    $E$.insert({$\bq,\bq_n$})
           </div>
+        return $V$, $E$
       </div>
       <hr/>
       def FindPath($V$, $E$, $\bq_s$, $\bq_t$): &emsp; <i>(online "query phase")</i>
@@ -297,8 +298,22 @@ <h1><a href="index.html" style="text-decoration:none;">Underactuated Robotics</a
 
     </example>
 
+    <subsection><h1>Getting smooth trajectories</h1>
+
+      <p>Even for kinematic motion planning, we might which to impose dynamic
+      constraints which require the path to be continuously differentiable (up to some
+      degree), and perhaps to respect velocity, acceleration, or other limits.</p>
+
+      <p>Post-processing PRM outputs.</p>
+
+      <p><a href="http://manipulation.mit.edu/trajectories.html#gcs">Kinematic
+      trajectory optimization w/ GCS</a> (using sampling to generate the regions).</p>
+    
+    </subsection>
+
   </section>
 
+
   <section><h1>Rapidly-exploring Random Trees (RRTs)</h1>
 
     <p>Thinking about how to extend the PRM to accommodate dynamic constraints leads up