The　pose　estimation　of　the　robot　depends　on　the　initial　spatial　parameters　of　the　camera　and　IMU　in　the　navigation　system.　However,　it　is　still　a　challenging　problem　to　calibrate　the　initial　spatial　parameters　online.　In　this　paper,　a　new　online　calibration　method　is　proposed　by　combining　the　advantages　of　Lie　group　and　optimal　attitude　estimation　calibration.　In　this　method,　the　Lie　group　is　used　to　represent　the　parameter　matrix　of　the　spatial　coordinate　relationship　of　the　sensor,　and　the　calibration　model　is　reconstructed　by　analyzing　the　influence　of　the　sensor　error　in　the　calculation　process.　Secondly,　the　lie　group　error　filter　is　studied　by　mapping　the　lie　group　error　equation　to　lie　algebraic　space　and　finite　field　expansion.　The　experimental　results　show　that　the　initial　alignment　method　has　higher　accuracy　and　better　robustness　than　the　existing　calibration　methods,　and　can　realize　the　online　calibration　of　the　camera　and　IMU　in　various　environments.