Role of reward shaping in object-goal navigation