Significant progress has been made in the navigation of autonomous mobile robots in indoor environments; however, poor map construction accuracy and poor path planning limit the practical applications of such robots. To solve these problems, a path planning algorithm based on guided search, the Gravitational Bidirectional Rapid Search Randomized Tree Algorithm (GBI-RRT) is proposed, which employs target bias sampling to efficiently guide nodes towards the target and reduce ineffective search. In order to further improve the navigation efficiency, another path reorganization strategy that eliminates low-quality nodes and improves the path curvature is proposed. It is integrated into a mobile robot based on a ROS system and evaluated in simulation and real environment experiments to verify the effectiveness of the above method. The results show that GBI-RRT outperforms the existing algorithms in various indoor environments.