increased joint strength to 40k

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

   m_Script: {fileID: 11500000, guid: 1b29724baddfa457da6eeab446fa49ca, type: 3}
   m_Name: 
   m_EditorClassIdentifier: 
-  maxJointSpring: 10000
-  jointDampen: 500
+  maxJointSpring: 40000
+  jointDampen: 3000
   maxJointForceLimit: 10000
   bodyPartsList: []
 --- !u!114 &8913670488868299308

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

         //reward looking at
 //        float facingReward = + 0.01f * Quaternion.Dot(m_OrientationCube.transform.rotation, hips.rotation)
 //                             + 0.01f * Quaternion.Dot(m_OrientationCube.transform.rotation, head.rotation);
-        float facingReward = + 0.1f * Quaternion.Dot(m_OrientationCube.transform.rotation, hips.rotation)
-                             + 0.1f * Quaternion.Dot(m_OrientationCube.transform.rotation, head.rotation);
-        float velReward = +0.2f * Vector3.Dot(m_OrientationCube.transform.forward,m_JdController.bodyPartsDict[hips].rb.velocity); //because we are observing in local space???
-        //Multiplying these amplifies the reward.
-        float runForwardTowardsTargetReward = facingReward * Mathf.Clamp(velReward, 0, 15);
+
+
+
+
+//        //Multiplying these amplifies the reward.
+//        float facingReward = + 0.1f * Quaternion.Dot(m_OrientationCube.transform.rotation, hips.rotation)
+//                             + 0.1f * Quaternion.Dot(m_OrientationCube.transform.rotation, head.rotation);
+//        float velReward = +0.2f * Vector3.Dot(m_OrientationCube.transform.forward,m_JdController.bodyPartsDict[hips].rb.velocity); //because we are observing in local space???
+//        float runForwardTowardsTargetReward = facingReward * Mathf.Clamp(velReward, 0, 15);
+        
+//        print(Quaternion.Angle(hips.transform.rotation, thighL.transform.rotation));
+
+
 //        print($"Combined {runForwardTowardsTargetReward}");
 //        float runBackwardsTowardsTargetReward = facingReward * Mathf.Clamp(velReward, -1, 0);
         // Set reward for this step according to mixture of the following elements.
         // d. Discourage head movement.
         AddReward(
-            runForwardTowardsTargetReward
+//            runForwardTowardsTargetReward
+//            + 0.02f * Vector3.Dot(m_OrientationCube.transform.forward,Vector3.ClampMagnitude(m_JdController.bodyPartsDict[hips].rb.velocity,5))
+            + 0.01f * Vector3.Dot(m_OrientationCube.transform.forward,Vector3.ClampMagnitude(m_JdController.bodyPartsDict[hips].rb.velocity,3))
-//            + 0.01f * (Quaternion.Dot(m_OrientationCube.transform.rotation, hips.rotation) - 1) * .5f //penalize not looking at
-//            + 0.01f * (Quaternion.Dot(m_OrientationCube.transform.rotation, head.rotation) - 1) * .5f //penalize not looking at
+            + 0.015f * (Quaternion.Dot(m_OrientationCube.transform.rotation, hips.rotation) - 1) * .5f //penalize not looking at
+            + 0.015f * (Quaternion.Dot(m_OrientationCube.transform.rotation, head.rotation) - 1) * .5f //penalize not looking at
-            + 0.02f * (head.position.y - hips.position.y)
-            - 0.01f * Vector3.Distance(m_JdController.bodyPartsDict[head].rb.velocity,
-                m_JdController.bodyPartsDict[hips].rb.velocity)
+            
+            
+//            + 0.02f * (head.position.y - hips.position.y)
+//            - 0.01f * Vector3.Distance(m_JdController.bodyPartsDict[head].rb.velocity,
+//                m_JdController.bodyPartsDict[hips].rb.velocity)
         );