The　development　of　an　automated　suturing　robot　for　oral　and　maxillofacial　surgery　(OMS)　has　long　been　a　challenging　task,　mainly　due　to　the　spatial　constraints　in　the　oral　cavity,　which　make　intraoral　navigation　difficult　to　realize.　In　this　paper,　we　propose　a　spatially　compact　visual　navigation　system　for　a　single-arm　suturing　robot　to　enable　real-time　high-precision　navigation　during　the　intraoperative　stage.　Correspondingly,　a　two-stage　intraoperative　navigation　framework　is　designed　to　achieve　intraoral　and　extra-oral　navigation.　By　designing　a　novel　mouth　opener　with　a　fixed　endoscope,　the　navigation　information　from　different　modules　can　be　aligned.　By　strategically　placing　flexible　position-sensitive　visual　markers　on　the　robot　body,　endeffector,　and　mouth　opener,　precise　pose　information　can　be　acquired　without　taking　up　additional　space,　thus　guiding　the　robot　to　automatically　perform　the　suturing　process.　In　addition,　we　design　a　normal　vector　guidance-bundle　adjustment　(NVG-BA)　module　and　a　dynamic　ROI　extraction　module　to　improve　the　system　performance.　We　constructed　a　real　suturing　robot　system　using　the　proposed　visual　navigation　framework　and　designed　a　large　number　of　localization　accuracy　experiments.　All　of　these　experiments　collectively　demonstrate　the　exceptional　localization　accuracy　and　safety　provided　by　the　proposed　navigation　system,　surpassing　previously　reported　OMS　navigation　systems,　all　while　avoiding　any　additional　occupation　of　intraoral　space.　Finally,　the　system　completed　an　automated　suturing　process　on　a　simulated　wound　of　a　head　phantom,　demonstrating　the　automated　suturing　capability　of　the　system.　IEEE