Objective In response to the issue of low precision of ultrasonic indoor positioning system in non-line-of-sight positioning a novel ultrasonic positioning system based on retroreflective ranging in non-line-of-sight environments was proposed starting from the overall positioning system to reduce errors in non-line-of-sight environments and hardware er sror such as clock synchronization. Methods Utilizing the advantages of the differential corrected Chan-Taylor algorithm combined with the Chan algorithm and Taylor series expansion algorithm the Chan-Taylor algorithm was used to estimate the known coordinate points in space and record their error information with actual coordinates as reference points. By differentially weighting unknown points with neighboring reference points within a certain range the initially estimated coordinates obtained by the Chan-Taylor algorithm were corrected to obtain the final position. In order to simplify the complexity of the positioning system and improve the positioning accuracy in line-of-sight environments an improved differential corrected Chan-Taylor algorithm was proposed. This involved reducing the initial density of reference points and marking the estimated coordinates of target points that meet the minimum interval conditions for reference points after differential correction as new reference points to optimize the distribution of error information in the original reference point system. Results Simulation experiments of the algorithm showed that in non-line-of-sight environments the average error in different reference point distribution areas of the differential corrected Chan-Taylor algorithm was reduced by 6. 43% to 37. 46% compared with the Chan algorithm and Chan-Taylor algorithm. The improved differential corrected Chan-Taylor algorithm reduced the average positioning error in line-of-sight positioning by at least 11. 15% with a decrease of 22. 59% in root mean square error. The positioning accuracy of the improved differential corrected algorithm was validated through the construction of an ultrasonic indoor positioning system with experimental results showing a positioning error range of 3 cm to 7. 5 cm where 90% of the errors were less than 6 cm representing a 28. 23% improvement compared with the Chan-Taylor algorithm. Conclusion The ultrasonic indoor positioning system shows a significant improvement in accuracy in non-line-of-sight positioning with a smaller improvement in line-of-sight positioning. The positioning algorithm can be optimized by improving the accuracy of the Chan-Taylor algorithm and enhancing the reference point weighting function. Hardware improvements such as optimizing the ultrasonic receiver signal recognition method and increasing the signal range at the transmission end can further enhance the positioning accuracy of the system.