Abstract
This paper describes an improved brain-inspired simultaneous localization and mapping (RatSLAM) that extracts visual features from saliency maps using a frequency-tuned (FT) model. In the traditional RatSLAM algorithm, the visual template feature is organized as a one-dimensional vector whose values only depend on pixel intensity; therefore, this feature is susceptible to changes in illumination intensity. In contrast to this approach, which directly generates visual templates from raw RGB images, we propose an FT model that converts RGB images into saliency maps to obtain visual templates. The visual templates extracted from the saliency maps contain more of the feature information contained within the original images. Our experimental results demonstrate that the accuracy of loop closure detection was improved, as measured by the number of loop closures detected by our method compared with the traditional RatSLAM system. We additionally verified that the proposed FT model-based visual templates improve the robustness of familiar visual scene identification by RatSLAM.
Highlights
In the past 30 years, traditional simultaneous localization and mapping (SLAM) algorithms based on probabilistic mathematical models, such as extended Kalman filter-based SLAM (EKF-SLAM) and Fast-SLAM (Thrun et al, 2005; Lv et al, 2014), have achieved remarkable results
We analyzed the experimental results from the perspectives of local view cell activity and of the experience maps generated by the traditional RatSLAM and FT model-based RatSLAM methods
The saliency model was applied to the RatSLAM system to extract visual templates, in order to improve the robustness of visual templates and the performance of the loop closure detection algorithm
Summary
In the past 30 years, traditional simultaneous localization and mapping (SLAM) algorithms based on probabilistic mathematical models, such as extended Kalman filter-based SLAM (EKF-SLAM) and Fast-SLAM (Thrun et al, 2005; Lv et al, 2014), have achieved remarkable results. The huge number of calculations, high complexity, and large mapping errors of these methods remain to be solved in the field of robot navigation. Compared with filter-based SLAM (Huang et al, 2013; Srivatsan et al, 2018) and optimizationbased SLAM (Hess et al, 2016), the “SLAM” performed in animals demonstrates perfect biological rationality and high adaptability to complex environments; for example, rodents such as mice have powerful navigational capabilities and can solve the entire SLAM problem, even in mazes with crossing paths. The RatSLAM algorithm based on rodents is widely accepted due to its strong biological rationality and low requirements for computing power (Milford et al, 2004; Milford and Wyeth, 2008; Yuan et al, 2015).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
