071b27b1ffc1c38e84f64fafcc126fafa54369a1,cistar-dev/cistar/envs/loop_accel.py,SimpleAccelerationEnvironment,compute_reward,#SimpleAccelerationEnvironment#Any#Any#,54

Before Change

            return max(max_cost - cost, 0)

        elif reward_type == "distance":
            distance = np.array([self.vehicles[veh_id]["absolute_position"] - self.initial_pos[veh_id]
                                 for veh_id in self.ids])

            return sum(distance)

    def getState(self):

After Change

        
        See parent class
        
        vel = state[0]

        if any(vel < -100) or kwargs["fail"]:
            return 0.0

        max_cost = np.array([self.env_params["target_velocity"]]*self.scenario.num_vehicles)

Instances