To　combine　the　advantages　of　humans　(adaptive　intelligence)　and　robots　(higher　movement　accuracy　and　physical　limitation),　we　propose　a　virtual　reality　human-robot　collaborative　welding　system　which　allows　them　to　collaborate　with　each　other　to　complete　welding　tasks.　A　common　welding　method,　weaving　gas　tungsten　arc　welding　(GTAW)　is　presented　as　a　case　study.　In　this　system,　a　6-degree　of　freedom　(6-DoF)　robot　UR-5　carrying　the　welding　torch　works　as　the　final　performer　of　weaving　welding.　A　virtual　welding　environment　based　on　HTC　VIVE　is　built　where　humans　can　observe　the　working　scene　via　a　motion-tracked　headset　without　being　onsite　physically.　A　mapping　model　between　arc　voltage,　arc　length　and　welding　current　is　established,　and　based　on　this　model　a　seam　tracking　algorithm　is　developed　which　allows　the　robot　to　track　the　weld　seam　automatically.　The　robot　is　pre-programmed　to　weave　the　welding　torch　across　the　weld　seam.　The　robot　travelling　along　the　weld　seam　is　fully　controlled　by　the　human　operator　via　motion-tracked　handle　per　the　observed　working　scene.　This　teleoperation　style　allows　the　humans　to　adjust　the　travel　speed　adaptively　and　freely　as　needed　without　suffering　onsite　danger.　The　proposed　collaborative　welding　system　combines　the　advantages　of　humans　and　robots　together,　utilizing　the　intelligence　and　adaptability　of　the　human　operator　with　precise　movement　obtained　from　robots.　The　welding　experiments　show　that　the　welded　workpiece　from　human-robot　collaboration　has　better　performance　compared　with　that　from　either　humans　or　robots　separately　and　demonstrated　the　effectiveness　of　the　proposed　virtual　reality　human-robot　collaborative　welding　system.