f5b25f57660910a0a2fdebc3d3a030e14172722c,softlearning/environments/gym/mujoco/pusher_2d_env.py,ForkReacherEnv,compute_reward,#ForkReacherEnv#Any#Any#,158

Before Change

            raise NotImplementedError("Might be broken.")

        arm_pos = observations[:, -6:-4]
        goal_pos = self.get_body_com("goal")[:2][None]
        object_pos = observations[:, -3:-1]

        arm_goal_dists = np.linalg.norm(arm_pos - goal_pos, axis=1)
        arm_object_dists = np.linalg.norm(arm_pos - object_pos, axis=1)

After Change

            raise NotImplementedError("Might be broken.")

        arm_pos = observations[:, -8:-6]
        goal_pos = observations[:, -2:]
        object_pos = observations[:, -5:-3]

        arm_goal_dists = np.linalg.norm(arm_pos - goal_pos, axis=1)
        arm_object_dists = np.linalg.norm(arm_pos - object_pos, axis=1)

Instances