The　control　of　permanent　magnet　synchronous　motor　is　one　of　the　core　technologies　in　robotics.　The　subject　not　only　requires　the　servo　motor　to　have　good　anti-interference　performance,　but　also　requires　the　servo　motor　to　be　able　to　better　track　the　position　and　follow-up　instructions　sent　by　the　central　processor.　Conventional　adaptive　disturbance　rejection　control　(ADRC)　mostly　studies　the　step　response　and　anti-interference　performance　of　the　motor,　but　the　tracking　performance　is　seldom　studied.　In　ADRC,　the　derivative　of　input　is　approximately　zero,　so　that　the　modeling　error　will　be　generated　by　a　time-varying　input　signal.　The　control　algorithm　cannot　observe　and　compensate　the　error　through　the　extended　state　observer,　so　the　tracking　error　of　the　motor　is　large.　An　improved　ADRC　is　proposed　to　solve　the　tracking　error　problem　of　conventional　ADRC.　The　theoretical　results　show　that　this　method　can　reduce　the　tracking　error　and　improve　the　bandwidth　of　the　motor.　Finally,　the　validity　of　the　method　is　verified　by　simulating　an　experiment.　©　Springer　Nature　Singapore　Pte　Ltd.　2020.