For　patients　with　ankle　injuries,　rehabilitation　training　is　an　important　and　effective　way　to　help　patients　restore　their　ankle　complex＇s　motor　abilities.　Aiming　to　improve　the　accuracy　and　performance　of　ankle　rehabilitation,　the　authors　focus　on　the　control　strategies　of　the　developed　parallel　ankle　rehabilitation　robot　with　novel　2-UPS/RRR　mechanism.　Firstly,　the　kinematics　model　of　the　mechanism　is　established,　and　they　deduce　the　inverse　solution　of　positions　as　well　as　the　velocity　mapping　between　the　driving　speed　and　the　robot＇s　angular　velocity,　based　on　which　they　realise　the　trajectory　tracking　control　in　the　process　of　passive　rehabilitation　training.　Secondly,　they　set　up　experiments　to　determine　the　torque　threshold　that　can　be　used　to　detect　the　motion　intention　of　ankle　joint,　and　then　they　propose　the　active　rehabilitation　training　strategy　according　to　the　motion　intention　detection.　Finally,　experiments　were　carried　out　with　healthy　subjects,　with　results　showing　that　the　trajectory　tracking　error　during　passive　rehabilitation　training　is　very　small,　and　the　moving　platform　of　the　ankle　rehabilitation　robot　can　drive　the　ankle　joint　to　the　detected　motion　intention　direction　at　a　constant　speed　flexibly　and　smoothly,　which　verifies　the　effectiveness　of　the　control　strategies　for　ankle　rehabilitation　training.　Copyright　©　2020　Cognitive　Computation　and　Systems.　All　rights　reserved.