This　paper　introduces　a　novel　state　and　input　constrained　optimal　tracking　control　method　to　address　limitations　posed　by　only　partially　known　robot　system　models　and　constrained　physical　variables,　such　as　joint　position,　velocity,　and　torque　during　the　control　process.　The　method　employs　the　slack　function　and　nonquadratic　function　methods　to　effectively　handle　state　error　and　input　constraints,　ensuring　the　overall　stationarity　and　safety　of　the　control　process.　Subsequently,　the　adaptive　dynamic　programming　(ADP)　algorithm　is　designed　to　formulate　update　laws　for　the　approximating　neural　networks,　enabling　the　derivation　of　optimal　control　solutions　without　relying　on　an　internal　dynamic　model　of　the　system.　Finally,　convergence　and　performance　are　validated　through　simulation　experiments,　confirming　its　efficacy　in　managing　constraints　and　demonstrating　its　application　perspectives　in　real-world　scenarios.　©　2024　IEEE.