Using　the　traditional　environmental　model　and　the　known　control　algorithm　can　lead　to　the　failure　of　the　task　easily　when　teleoperation　robot　works　in　unstructured　unknown　environment,　because　of　the　lack　of　information　about　the　environment　around　the　robot　arm　and　the　lack　of　feedback.　To　solve　this　significant　problem,　a　new　improved　algorithm　based　on　the　traditional　approximate　obstacle　avoidance　algorithm　is　proposed　in　this　paper.　By　adding　control　algorithm　factors,　this　algorithm　will　complete　the　obstacle　avoidance　in　unknown　environment　through　transforming　information　which　is　obtained　from　the　obstacle　distance　detector　into　each　joint　and　the　velocity　parameter　of　the　end　effector　and　mapping　to　the　joint　movement　by　Jacobian　matrix.　Thus,　the　performance　of　the　robot　arm　is　improved.　In　addition,　the　algorithm　has　the　advantages　of　small　computation　and　fast　escape　speed.　Finally,　the　simulation　experiment　of　obstacle　avoidance　is　carried　out　in　MATLAB　software　to　verify　the　effectiveness　of　the　algorithm.　The　results　show　that　the　algorithm　can　accomplish　the　obstacle　avoidance　of　teleoperation　manipulator　in　real-time　and　effectively　in　unstructured　unknown　environment.