This　article　is　inspired　by　the　firing　mechanism　of　hippocampus　place　cells　to　build　an　environmental　cognitive　map,　and　introduces　a　Successor　Representation　(SR)　model　to　realize　robot　path　planning.　Compared　to　the　RL　method　based　on　value　learning　used　in　current　brain　navigation　models,　this　model　does　not　rely　on　specific　goal　points　for　learning　cognitive　maps.　This　paper　conducted　experiments　on　the　model　in　a　maze　environment,　and　the　results　showed　that　the　model　can　achieve　fast　switching　between　goal　points　and　plan　the　optimal　path　without　retraining　the　value　function,　with　high　flexibility.　©　2023　IEEE.