Pathfinding　is　widely　applied　when　encountering　autonomous　driving,　mobile　robot　pathfinding,　and　so　on.　Traditional　pathfinding　algorithms　have　certain　limitations　such　as　high　computational　cost,　overly　dependent　on　the　design　of　heuristics,　and　unable　to　focus　on　muti-objective.　This　research　deals　with　the　comparison　of　three　algorithms　applied　in　the　same　pathfinding　situation.　A　delivery　robot　is　taken　as　the　object　of　study,　and　the　simulation　of　the　same　map　and　obstacle　design　ensures　three　algorithms　are　tested　under　the　same　circumstances.　To　improve　the　A　star　algorithm,　enhancements　are　applied,　such　as　real-time　heuristic　adjustments,　adaptive　cost　functions,　and　dynamic　re-planning　techniques.　These　modifications　allow　the　algorithm　to　efficiently　re-evaluate　paths　as　environmental　conditions　evolve,　ensuring　timely　and　optimal　decisionmaking.　After　using　Python　coding　to　manipulate　the　simulation　of　A　star,　Dijkstra,　and　rapidly　exploring　random　tree　(RRT)　algorithms,　A　star　algorithm　performs　best　in　the　pathfinding　problems.　Research　in　the　A　star　algorithm　enables　significant　improvement　in　the　calculation　efficiency　by　enhancing　the　flexibility　of　the　heuristic　function　used　in　A　star　while　cutting　back　on　search　space　and　computation　time　spent　which　leads　to　a　great　improvement　in　pathfinding　tasks.　©　The　Institution　of　Engineering　&　Technology　2024.