In　order　to　protect　joints　from　impact　damage　during　the　process　of　space　robot　capturing　satellite,　a　springdamper　buffer　device　is　designed　between　joint　motor　and　manipulator.　The　device　can　not　only　use　spring　to　achieve　the　compliance　during　the　capture　operation,　use　damper　to　absorb　impact　energy　and　suppress　flexible　vibration;　but　also　limit　the　joint＇s　impact　torque　to　a　safe　range　through　reasonable　and　coordinated　design　the　collision　avoidance　and　compliance　strategy.　First　of　all,　the　dynamic　models　of　space　robot　and　satellite　at　collision　time　are　derived　by　using　Lagrange　function　based　on　dissipation　theory　and　Newton-Euler　function　respectively.　After　that,　combined　with　Newton＇s　third　law,　velocity　and　position　constraints　of　capture　points　constraints　of　bodies,　the　dynamic　model　of　hybrid　system　after　capture　is　obtained,　the　impact　effect　and　impact　force　are　calculated　based　on　momentum　conservation.　Finally,　a　collision　avoidance　and　compliance　reinforcement　learning　control　strategy　with　buffer　device　is　proposed.　The　penalty　signal　is　obtained　by　trial-and-error　interaction　with　dynamic　environment,　be　used　to　optimize　the　controller　to　stabilize　the　instability　hybrid　system.　The　stability　of　the　system　is　proved　by　Lyapunov　theorem,　and　the　impact　resistance　of　the　device　and　the　effectiveness　of　the　proposed　strategy　are　verified　by　numerical　simulation.　©　2020,　Editorial　Department　of　Control　Theory　&　Applications　South　China　University　of　Technology.　All　right　reserved.