[MLA-1223] Backport Heuristic fixes (#4176)

* [bugfix] Make FoodCollector heuristic playable (#4147)
* Modified the documentation of the Heuristic method (default action = previous action) (#4174)

Co-authored-by: Ervin T <[email protected]>
Co-authored-by: Vincent-Pierre BERGES <[email protected]>

Author: 3 people

Project/Assets/ML-Agents/Examples/FoodCollector/Scripts/FoodCollectorAgent.cs

@@ -209,6 +209,9 @@ public override void OnActionReceived(float[] vectorAction)
 
     public override void Heuristic(float[] actionsOut)
     {
+        actionsOut[0] = 0f;
+        actionsOut[1] = 0f;
+        actionsOut[2] = 0f;
         if (Input.GetKey(KeyCode.D))
         {
             actionsOut[2] = 2f;

Project/Assets/ML-Agents/Examples/Soccer/Scripts/AgentSoccer.cs

@@ -166,6 +166,7 @@ public override void OnActionReceived(float[] vectorAction)
 
     public override void Heuristic(float[] actionsOut)
     {
+        Array.Clear(actionsOut, 0, actionsOut.Length);
         //forward
         if (Input.GetKey(KeyCode.W))
         {

Project/Assets/ML-Agents/Examples/WallJump/Scripts/WallJumpAgent.cs

@@ -242,6 +242,7 @@ public override void OnActionReceived(float[] vectorAction)
 
     public override void Heuristic(float[] actionsOut)
     {
+        System.Array.Clear(actionsOut, 0, actionsOut.Length);
         if (Input.GetKey(KeyCode.D))
         {
             actionsOut[1] = 2f;

com.unity.ml-agents/CHANGELOG.md

@@ -12,6 +12,8 @@ and this project adheres to
 - Fixed an issue where RayPerceptionSensor would raise an exception when the
 list of tags was empty, or a tag in the list was invalid (unknown, null, or
 empty string). (#4155)
+#### ml-agents / ml-agents-envs / gym-unity (Python)
+- Fixed issue with FoodCollector, Soccer, and WallJump when playing with keyboard. (#4147, #4174)
 
 ## [1.0.2] - 2020-06-04
 ### Minor Changes

com.unity.ml-agents/Runtime/Agent.cs

@@ -761,6 +761,8 @@ public virtual void Initialize() {}
         ///
         /// Your heuristic implementation can use any decision making logic you specify. Assign decision
         /// values to the float[] array, <paramref name="actionsOut"/>, passed to your function as a parameter.
+        /// The same array will be reused between steps. It is up to the user to initialize
+        /// the values on each call, for example by calling `Array.Clear(actionsOut, 0, actionsOut.Length);`.
         /// Add values to the array at the same indexes as they are used in your
         /// <seealso cref="OnActionReceived(float[])"/> function, which receives this array and
         /// implements the corresponding agent behavior. See [Actions] for more information