How　to　make　robots　explore　effectively　in　unknown　environments　is　one　of　the　key　issues　in　robot　navigation.　Among　them,　the　exploration　of　efficiency　and　traversal　issues　is　the　embodiment　of　effectiveness.　To　solve　this　problem,　an　Intelligent　Rapidly-exploration　Random　Tree　(IRRT)　algorithm　based　on　evolutionary　cognition　is　proposed.　This　algorithm　can　intelligently　adjust　the　growth　rate　of　the　global　tree　according　to　the　complete　situation　of　the　exploration　map.　Using　the　method　of　simultaneous　generation　of　global　tree　and　local　tree　to　explore　boundary　points,　the　detected　boundary　points　are　filtered　by　the　DBSCAN　clustering　method　to　obtain　the　target　point　set,　which　is　used　as　the　current　target　point　of　the　robot　movement.　Simulation　verification　in　the　Robot　Operating　System　(ROS)　shows　that　the　algorithm　can　complete　the　traversal　of　the　unknown　environment,　and　effectively　reduce　the　time　cost.　©　2020　IEEE.