tune

Author: sshane

selfdrive/controls/lib/longitudinal_mpc_lib/long_mpc.py

@@ -35,8 +35,8 @@
 X_EGO_COST = 0.
 V_EGO_COST = 0.
 A_EGO_COST = 0.
-J_EGO_COST = 5.0
-A_CHANGE_COST = .5
+J_EGO_COST = 4.0
+A_CHANGE_COST = .4
 DANGER_ZONE_COST = 100.
 CRASH_DISTANCE = .5
 LIMIT_COST = 1e6
@@ -246,8 +246,8 @@ def set_weights_for_lead_policy(self, accelerating):
     d_zone_cost_multiplier = 1  # interp(self.desired_TR, TRs, [4., 1.0, 1.0])
     _J_EGO_COST, _A_CHANGE_COST = J_EGO_COST, A_CHANGE_COST
     if accelerating:
-      _J_EGO_COST = J_EGO_COST * interp(self.v_ego, [2.2, 5.0], [0.0, 1.0])
-      _A_CHANGE_COST = A_CHANGE_COST * interp(self.v_ego, [2.2, 5.0], [0.0, 1.0])
+      _J_EGO_COST = J_EGO_COST * interp(self.v_ego, [2.0, 8.0], [0.1, 1.0])
+      _A_CHANGE_COST = A_CHANGE_COST * interp(self.v_ego, [2.0, 8.0], [0.1, 1.0])
 
     W = np.asfortranarray(np.diag([X_EGO_OBSTACLE_COST * x_ego_obstacle_cost_multiplier, X_EGO_COST, V_EGO_COST, A_EGO_COST, _A_CHANGE_COST, _J_EGO_COST]))
     for i in range(N):