Collaborative　learning　is　widely　applied　in　practical　education　due　to　its　high　efficiency　and　positive　effectiveness.　Virtual　reality　(VR)　has　driven　the　development　of　collaborative　learning　with　its　immersive　and　interactive　characteristics.　Task　allocation,　as　the　first　step　in　initiating　collaboration　when　applying　VR　to　collaborative　learning,　plays　a　crucial　role.　However,　it　has　not　received　enough　attention　and　lacks　in-depth　research.　Therefore,　we　first　analyzed　the　characteristics　of　VR　collaborative　learning　and　found　that　considering　the　influence　of　five　key　factors　on　tasks　allocation　can　improve　learning　outcomes.　Consequently,　we　constructed　a　model　applicable　to　VR　collaborative　learning.　Subsequently,　we　proposed　a　two-way　task　allocation　strategy　that　balances　the　interaction　between　learners＇　intentions　and　the　requirements　of　tasks.　Then,　by　invoking　an　improved　sparrow　search　algorithm　for　optimization　calculations,　the　assignment　results　were　automatically　computed　after　going　through　three　stages.　Finally,　evaluation　experiments　were　conducted　to　validate　the　feasibility　and　correctness　of　our　model　and　strategy　by　comparing　them　with　other　methods.　The　results　indicate　that　compared　to　conventional　collaborative　learning,　VR　collaborative　learning　yields　better　learning　outcomes.　Moreover,　our　strategy　demonstrates　higher　learning　efficiency　and　better　learning　effects　compared　to　using　self-negotiation　and　Agent-based　strategies　for　assigning　VR　collaborative　tasks,　providing　a　valuable　reference　for　the　continuous　exploration　of　this　scientific　issue.