Update README.md

Author: verityw

README.md

@@ -88,7 +88,8 @@ How you implement the wall follower is entirely up to you. However implementing
 * __Set up ROS structure__: Set up your wall follower node so that it subscribes to laser messages and publishes steering commands. Make sure you can at least make the racecar move fowards at a constant speed and turning angle before working on your controller.
 * __Slice up the scan__: Consider slicing the ```ranges``` data into more useful pieces. A majority of the data won’t be useful to you if you only care about a wall to one side. When you can, try to use [```numpy```](https://numpy.org/) operations rather than for loops in your code. [Multidimensional slicing](https://docs.scipy.org/doc/numpy-1.13.0/reference/arrays.indexing.html) and [broadcasting](https://docs.scipy.org/doc/numpy-1.13.0/user/basics.broadcasting.html) can make your code cleaner and much more efficient. You can turn any array into a ```numpy``` array with [```np.array```](https://docs.scipy.org/doc/numpy-1.13.0/reference/generated/numpy.array.html), or you can integrate it directly with ros like in [this tutorial](http://wiki.ros.org/rospy_tutorials/Tutorials/numpy).
 * __Find the wall__: There are many ways to detect a wall in a laser scan. In a perfect world you might be able to detect it using a single sample of the LIDAR data. However with noisy data and uneven surfaces this might not be enough. A [least squares regression](https://en.wikipedia.org/wiki/Simple_linear_regression) is an easy way to account for more noise. The [RANSAC](https://en.wikipedia.org/wiki/Random_sample_consensus) algorithm can “upgrade” an existing model (like least squares) to be more robust to outliers. _Note: Attempt RANSAC only if you've already built a functional wall follower. It is probably overkill._
-* __Use PD or PID control__: A robust wall follower algorithm that can handle wavy wall contours and corners should probably use some sort of [PD or PID control](https://en.wikipedia.org/wiki/PID_controller#Discrete_implementation). Simple P (proportional) control is often not enough to create a responsive and stable system.
+* __Use PD or PID control__: A robust wall follower algorithm that can handle wavy wall contours and corners should probably use some sort of [PD or PID control](https://en.wikipedia.org/wiki/PID_controller#Discrete_implementation). Simple P (proportional) control is often not enough to create a responsive and stable sy--stem.
+* __Use the visualization code__: We provided an example Python script in `src` that plots a line in Rviz. Try to use this to make sure your code (e.g. wall detection) is working!
 
 ## Starter Code