Abstract: Rat-simultaneous localization and mapping (RatSLAM) is a brain-inspired navigation model that simulates the neural processing mechanisms of navigation in the rodent brain. Unfortunately, the existing RatSLAM systems primarily rely on passive datasets for experimentation and map building. This passive SLAM approach generally involves collecting environmental data through preset paths or manually controlling the robot, which often limits its performance in new environments.To overcome this limitation, this paper reports the proposal of an improved autonomous exploration strategy based on the rapidly exploring random tree (RRT) algorithm for active environmental exploration. The existing mainstream autonomous exploration algorithms are generally characterized by their insufficient coverage of the internal areas of target environments and may require frequent path adjustments in dynamic environments, which is not conducive to the localization and map updating performances of RatSLAM. Conversely, conventional RRT algorithm-based autonomous exploration strategies can efficiently explore high-dimensional, complex environments, adapt to dynamic changes in real time, and provide comprehensive environmental data. However, the paths generated by these exploration methods are not always sufficiently smooth, do not conform to the regularities of exploring animals' environments, and cannot readily simulate the real movement trajectories of animals, thus failing to satisfy the data collection requirements of brain-inspired navigation. Therefore, an improved autonomous exploration strategy is proposed in this study based on the integration of the RRT algorithm with the RatSLAM system. This strategy enables the autonomous movement of a robot to collect environmental data in unknown environments, whereas the integrated RatSLAM system processes these data to generate a cognitive map of the target environment. Finally, the results obtained from simulation and real-world experiments reveal that the proposed method achieves good results in active map building, similar to how biological entities explore environments, thereby validating the feasibility of the algorithm.