As　the＇blood　vessel＇　of　the　city,　pipelines　are　all　over　the　underground　of　the　city.　To　inspect　the　pipeline　conveniently　and　intuitively,　the　pipeline　robot　needs　to　enter　the　pipeline　and　transmit　the　real-time　image　inside　the　pipeline　through　the　camera　module　of　the　robot.　However,　most　pipeline　robots　rely　on　manual　attitude　control　for　turning　and　other　operations.　This　paper　studies　the　problem　of　automatic　navigation　when　the　robot　is　running　in　the　pipeline　with　high　water　pressure,　which　can　assist　the　operator　to　control　the　robot.The　basic　principle　is　that　the　video　signal　of　the　camera　is　transmitted　to　the　host　computer　through　the　optical　fiber,　and　the　upper　computer　processes　the　image　to　determine　whether　to　turn　and　the　direction　of　turning,　and　then　sends　instructions　to　the　attitude　change　module.　This　method　can　accurately　judge　the　turning　direction,　and　good　results　are　obtained　in　the　experiment.　©　2021　IEEE