Added explanation of scaled trajectory controller (UniversalRobots#250)

* Added explanation of scaled trajectory controller

* Apply suggestions from code review

Thanks @gavanderhoorn for this in-depth review!

Co-authored-by: G.A. vd. Hoorn <[email protected]>

* Added another suggestion I missed

Co-authored-by: G.A. vd. Hoorn <[email protected]>

* Clarified where the speed_scaling value comes from.

* Format ros_control correctly

* Rewrote bullet point about non-moving robot

* Replace time frame with control cycle

Co-authored-by: G.A. vd. Hoorn <[email protected]>

Author: Felix Exner

ur_controllers/README.md

@@ -1,18 +1,76 @@
 # ur_controllers
 
-This package contains controllers and hardware interface for ROS control that are special to the UR
+This package contains controllers and hardware interface for `ros_control` that are special to the UR
 robot family. Currently this contains
 
   * A **speed_scaling_interface** to read the value of the current speed scaling into controllers.
   * A **scaled_joint_command_interface** that provides access to joint values and commands in 
   combination with the speed scaling value.
-  * A **speed_scaling_state_controller** that publishes the current value of the speed scaling
-  to a topic interface.
+  * A **speed_scaling_state_controller** that publishes the current execution speed as reported by
+  the robot to a topic interface. Values are floating points between 0 and 1.
   * A **scaled_joint_trajectory_controller** that is similar to the *joint_trajectory_controller*,
   but it uses the speed scaling reported by the robot to reduce progress in the trajectory.
 
 ## About this package
-This package contains controllers not being available in the default ROS-Control set. They are
+This package contains controllers not being available in the default `ros_control` set. They are
 created to support more features offered by the UR robot family. Any of these controllers are
 example implementations for certain features and are intended to be generalized and merged
-into the default ROS-Control controller set at some future point.
+into the default `ros_control` controller set at some future point.
+
+## Controller description
+This packages offers a couple of specific controllers that will be explained in the following
+sections.
+### ur_controllers/SpeedScalingStateController
+This controller publishes the current actual execution speed as reported by the robot. Values are
+floating points between 0 and 1.
+
+In the [`ur_robot_driver`](../ur_robot_driver) this is calculated by multiplying the two [RTDE](https://www.universal-robots.com/articles/ur/real-time-data-exchange-rtde-guide/) data
+fields `speed_scaling` (which should be equal to the value shown by the speed slider position on the
+teach pendant) and `target_speed_fraction` (Which is the fraction to which execution gets slowed
+down by the controller).
+### position_controllers/ScaledJointTrajectoryController and velocity_controllers/ScaledJointTrajectoryController
+These controllers work similar to the well-known
+[`joint_trajectory_controller`](http://wiki.ros.org/joint_trajectory_controller).
+
+However, they are extended to handle the robot's execution speed specifically. Because the default
+`joint_trajectory_controller` would interpolate the trajectory with the configured time constraints (ie: always assume maximum velocity and acceleration supported by the robot),
+this could lead to significant path deviation due to multiple reasons:
+ - The speed slider on the robot might not be at 100%, so motion commands sent from ROS would
+   effectively get scaled down resulting in a slower execution.
+ - The robot could scale down motions based on configured safety limits resulting in a slower motion
+   than expected and therefore not reaching the desired target in a control cycle.
+ - Motions might not be executed at all, e.g. because the robot is E-stopped or in a protective stop
+ - Motion commands sent to the robot might not be interpreted, e.g. because there is no
+   [`external_control`](https://github.com/UniversalRobots/Universal_Robots_ROS_Driver#prepare-the-robot)
+   program node running on the robot controller.
+ - The program interpreting motion commands could be paused.
+
+The following plot illustrates the problem:
+![Trajectory execution with default trajectory controller](doc/traj_without_speed_scaling.png
+"Trajectory execution with default trajectory controller")
+
+The graph shows a trajectory with one joint being moved to a target point and back to its starting
+point. As the joint's speed is limited to a very low setting on the teach pendant, speed scaling
+(black line) activates and limits the joint speed (green line). As a result, the target
+trajectory (light blue) doesn't get executed by the robot, but instead the pink trajectory is executed.
+The vertical distance between the light blue line and the pink line is the path error in each
+control cycle. We can see that the path deviation gets above 300 degrees at some point and the
+target point at -6 radians never gets reached.
+
+All of the cases mentioned above are addressed by the scaled trajectory versions. Trajectory execution
+can be transparently scaled down using the speed slider on the teach pendant without leading to
+additional path deviations. Pausing the program or hitting the E-stop effectively leads to
+`speed_scaling` being 0 meaning the trajectory will not be continued until the program is continued.
+This way, trajectory executions can be explicitly paused and continued.
+
+With the scaled version of the trajectory controller the example motion shown in the previous diagram becomes:
+![Trajectory execution with scaled_joint_trajectory_controller](doc/traj_with_speed_scaling.png
+"Trajectory execution with scaled_joint_trajectory_controller")
+
+The deviation between trajectory interpolation on the ROS side and actual robot execution stays minimal and the
+robot reaches the intermediate setpoint instead of returning "too early" as in the example above.
+
+Under the hood this is implemented by proceeding the trajectory not by a full time step but only by
+the fraction determined by the current speed scaling. If speed scaling is currently at 50% then
+interpolation of the current control cycle will start half a time step after the beginning of the
+previous control cycle.