add dir vector obsv

Project/Assets/ML-Agents/Examples/Walker/Prefabs/DynamicPlatformWalker.prefab

       value: 
       objectReference: {fileID: 11400000, guid: 465050f5bacbb4ec285a0d912a0f4f67,
         type: 3}
+    - target: {fileID: 895268871377934297, guid: 765582efd9dda46ed98564603316353f,
+        type: 3}
+      propertyPath: m_BrainParameters.VectorObservationSize
+      value: 236
+      objectReference: {fileID: 0}
     - target: {fileID: 895268871377934298, guid: 765582efd9dda46ed98564603316353f,
         type: 3}
       propertyPath: m_LocalPosition.x

Project/Assets/ML-Agents/Examples/Walker/Scenes/WalkerDynamic.unity

       propertyPath: m_LocalPosition.x
       value: 0
       objectReference: {fileID: 0}
+    - target: {fileID: 6065910098925129095, guid: f51e8260728fd4c8fa87bcda9d0e2027,
+        type: 3}
+      propertyPath: m_BrainParameters.VectorObservationSize
+      value: 236
+      objectReference: {fileID: 0}
     m_RemovedComponents: []
   m_SourcePrefab: {fileID: 100100000, guid: f51e8260728fd4c8fa87bcda9d0e2027, type: 3}
 --- !u!1001 &3766798870867890383

Project/Assets/ML-Agents/Examples/Walker/Scripts/WalkerAgent.cs

         sensor.AddObservation(m_OrientationCube.transform.InverseTransformDirection(bp.rb.angularVelocity));
         
         //Get position relative to hips in the context of our orientation cube's space
-        sensor.AddObservation(m_OrientationCube.transform.InverseTransformDirection(bp.rb.position - hips.position));  //best
+        sensor.AddObservation(m_OrientationCube.transform.InverseTransformDirection(bp.rb.position - hips.position));
 
         if (bp.rb.transform != hips && bp.rb.transform != handL && bp.rb.transform != handR)
         {
     {
         sensor.AddObservation(Quaternion.FromToRotation(hips.forward, m_OrientationCube.transform.forward));
         sensor.AddObservation(Quaternion.FromToRotation(head.forward, m_OrientationCube.transform.forward));
+        
+        //clamp the distance vector in case the target is far away. normalized to 1.
+//        var clampedDistFromCubeToTarget = (Vector3.ClampMagnitude(m_OrientationCube.transform.InverseTransformPoint(target.position), 15))/15;
+//        sensor.AddObservation(m_OrientationCube.transform.InverseTransformPoint(clampedDistFromCubeToTarget));
+        sensor.AddObservation(m_OrientationCube.transform.InverseTransformPoint(target.position));
+
+//        print(m_OrientationCube.transform.InverseTransformPoint(target.position));
+//        sensor.AddObservation(target.position - m_OrientationCube.transform.position);
 
         foreach (var bodyPart in m_JdController.bodyPartsList)
         {

config/trainer_config.yaml

     time_horizon: 1000
     batch_size: 2048
     buffer_size: 20480
-    # max_steps: 2e7
-    max_steps: 1e8
+    max_steps: 2e7
+    # max_steps: 1e8
     summary_freq: 30000
     num_layers: 3
     hidden_units: 512

Project/Assets/ML-Agents/Examples/Walker/TFModels/brandonh_2020-05-16_2020-05-16-19-55-55_walkdy1e8-ppo_results_walkdy1e8-ppo_WalkerDynamic.nn.meta

+fileFormatVersion: 2
+guid: b84f979c828524db3937219508446a7a
+ScriptedImporter:
+  fileIDToRecycleName:
+    11400000: main obj
+    11400002: model data
+  externalObjects: {}
+  userData: 
+  assetBundleName: 
+  assetBundleVariant: 
+  script: {fileID: 11500000, guid: 19ed1486aa27d4903b34839f37b8f69f, type: 3}