A　new　hybrid　index　dynamic　obstacle　avoidance　algorithm　for　redundant　manipulator　is　proposed.　Considering　the　randomness　of　the　moving　obstacle,　the　shortest　distance　index　is　used　as　the　index　of　the　obstacle　avoidance　planning　based　on　gradient　projection　method　when　the　moving　obstacle　is　inside　the　configuration　of　the　manipulator.　When　the　moving　obstacle　is　outside　the　configuration,　the　obstacle　avoidance　area　index　is　used　to　substitute　for　the　invalid　shortest　distance　index.　Through　using　the　hybrid　index,　the　combination　of　fine　adjustment　and　coarse　adjustment　for　the　configuration　of　the　manipulator　is　realized.　And　the　effect　of　the　whole　obstacle　avoidance　is　improved.　In　addition,　the　problem　of　exceeding　the　joint　velocity　limit　can　be　solved　by　the　on-line　adjustment　of　the　magnification　factor,　so　as　to　keep　a　constant　ratio　between　null-space-solution　and　minimum-norm-solution.　The　algorithm　is　illustrated　by　a　planar　3R2P　redundant　manipulator.　The　simulations　of　comparison　between　single　index　planning　and　hybrid　index　planning　show　superiority　and　effectiveness　of　the　latter.　And　the　real　object　experiments　prove　the　real-time　performance　and　universality　of　the　algorithm.　©　2010　Journal　of　Mechanical　Engineering.