Manipulator　path　planning　refers　to　planning　the　movement　trajectory　of　a　manipulator　through　algorithms　and　techniques　to　realize　accurate　and　efficient　operation.　It　is　widely　used　in　industrial　production　lines,　medical　surgeries,　and　other　fields.　The　sampling-based　rapid　exploration　random　tree　(RRT)　algorithm　has　been　widely　used　in　high-dimensional　manipulators　due　to　its　probabilistic　completeness　and　outstanding　expandability.　However,　the　algorithm　is　purely　random　in　the　sampling　session,　which　leads　to　the　existing　RRT　planning　algorithms　still　having　low　planning　efficiency　and　poor　path　quality.　To　solve　these　problems,　this　paper　proposes　a　new　and　improved　method,　which　assigns　a　cost　value　to　the　sampling　points　and　bases　the　selection　of　sampling　points　on　the　cost　value,　so　that　the　growth　of　the　path　tree　can　be　oriented　to　the　goal　point　more　quickly.　A　new　evaluation　metric,　i.e.,　the　average　angle　of　each　node　in　the　path,　is　also　designed　to　evaluate　the　quality　of　the　path,　and　the　smoother　the　path　is,　the　more　the　metric　converges　to　180°.　By　comparing　our　improved　RRT　algorithm　with　the　original　algorithm　in　several　different　scenarios,　the　results　show　that　our　improvement　possesses　shorter　running　time,　shorter　path　length,　and　larger　average　angle　of　path.　©　2024　Technical　Committee　on　Control　Theory,　Chinese　Association　of　Automation.