Welding　is　a　skill-demanding,　labor-intensive　operation　which　should　be　automated　whenever　possible.　Unfortunately,　manual　welding　will　still　be　irreplaceable　in　applications　where　the　needed　versatility　and　accessibility　cannot　be　achieved　by　robots.　On　the　other　hand,　the　skills　needed　for　manual　welding　typically　require　a　long　time　to　develop.　In　particular,　maintaining　the　torch　to　travel　in　desired　speed　is　a　challenge.　This　paper　thus　proposes　to　use　a　feedback　control　system　to　assist　the　welder　in　regulating　his/her　arm　movement　in　real　time　to　achieve　the　desired　torch　travel　speed.　To　this　end,　the　travel　speed　(system　output)　is　tracked　by　a　motion　sensor.　The　welder　is　instructed　by　a　visual　command　to　adjust　the　travel　speed　as　system　input.　An　auto-regressive　moving　average　model　has　been　used　to　correlate　the　system　output　to　the　system　input　as　the　human　welder　response　model.　To　determine　how　the　input　needs　to　be　adjusted　to　assisting　the　human　welder　to　achieve　the　desired　speed,　a　feedback　control　algorithm　has　been　developed.　Experiments　verified　that　the　proposed　feedback　control　system　is　capable　of　assisting　the　human　welder　to　track　the　desired　travel　speed.