Toward understanding natural language directions | IEEE Conference Publication | IEEE Xplore