Explaining Autonomous Driving Actions with Visual Question Answering