This　work　proposes　a　synchronous　obstacle　avoidance　tracking　control　framework　with　state　constraints　and　interference　compensation　to　achieve　autonomous　obstacle　avoidance　and　track　reference　paths　for　snake　robots.　In　the　kinematic,　an　improved　robust　line-of-sight　(LOS)　strategy　based　on　reduced-order　extended　state　observer　is　proposed.　The　sideslip　angle　estimation　compensates　for　the　deviation　of　the　direction　angle,　thereby　ensuring　the　accuracy　of　direction　commands.　Meanwhile,　considering　obstacle　avoidance　and　non-differentiable　inflection　points　of　fold　paths,　a　virtual　dynamic　guidance　rule　is　designed　to　generate　smooth　references,　which　reduces　the　computational　load　of　an　actual　robot　turning　and　obstacle　avoidance　processes.　In　the　dynamic,　a　barrier　Lyapunov-based　neural　network　robust　controller　is　derived　to　cope　with　state　constraints　caused　by　actuator　saturation.　The　adaptive　law　of　dynamic　deviation　and　external　interference　compensates　for　joint　torque.　Robots＇　autonomous　obstacle　avoidance　ability　and　the　superiority　of　the　proposed　method　are　demonstrated　through　experiments.