The　radial　artery　pulse　signal　contains　rich　information　about　human　health　status　and　diseases.　Currently,　most　acquisition　systems　take　the　form　of　fixed　acquisition　positions　that　cannot　automatically　locate　and　collect　the　pulse　and　require　manual　assistance.　However,　due　to　variations　in　human　wrist　anatomy,　these　systems　face　challenges　in　ensuring　consistent　and　repeatable　positioning　for　each　pulse　collection,　which　may　affect　the　quality　of　pulse　signals.　To　overcome　these　limitations,　this　paper　develops　an　automatic　pulse　signal　acquisition　system　based　on　a　combined　visual　and　tactile　pulse-finding　algorithm.　It　can　identify　the　strongest　pulse　position　through　a　lightweight　pulse　localization　network　(LPLN)　and　automatically　acquire　pulse　signals.　Additionally,　the　system　can　further　refine　the　identification　of　the　strongest　pulse　position　through　a　tactile　pulse-finding　method　based　on　time　domain　and　frequency　domain　characteristics.　The　adjusted　position　can　serve　as　training　data　for　LPLN　self-learning.　The　experimental　results　show　that　the　average　error　of　LPLN　visual　and　tactile　pulse-finding　algorithm　localization　is　4.07　mm　and　1.36　mm,　respectively.　The　proposed　prototype　may　serve　as　a　valuable　tool　for　intelligent　pulse　signal　acquisition,　guaranteeing　accurate　location　and　signal　quality.　IEEE