Automatic　parking　is　a　common　but　very　important　function　in　the　autonomous　driving　of　intelligent　vehicles.　At　the　same　time,　with　the　development　of　artificial　intelligence,　Deep　Reinforcement　Learning　has　become　a　popular　learning　algorithm　in　the　field　of　unmanned　autonomous　systems　with　its　strong　learning　ability　and　highly　intelligent　characteristics.　In　this　paper,　Deep　Reinforcement　Learning　algorithm　is　used　to　realize　automatic　parking,　and　problems　existing　in　parking　tasks　are　analyzed.　Based　on　the　dynamic　adjustment　of　the　agent’s　exploration　ability,　an　adaptive　network　parameter　noise　mechanism　is　proposed.　Compared　with　the　traditional　method　of　adding　noise　to　the　agent’s　output　actions,　this　mechanism　can　dynamically　adjust　the　variance　of　the　added　noise,　and　then　boost　the　agent’s　exploration　ability.　By　combining　with　TD3,　a　novel　algorithm—Adaptive　Parameter　Noise-TD3　(A-Noise-TD3)　is　developed.　Through　the　analysis　of　simulation　results,　the　effectiveness　of　the　proposed　method　is　verified.　©　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.　2025.