For high-precision positioning applications in large-scale urban environment, since the performance of traditional stand-alone GNSS-based RTK degrades greatly due to the signal blockage and multipath effect, multi-sensor integration systems become the most popular solutions most of which require to build LiDAR maps before positioning. However, not all the applications are suited to this scheme. For the applications that cannot build prior maps, if only odometry-based sensors are used, the issue of accumulative error will arise in large-scale area. Therefore, we still hope to use RTK in multi-sensor integration to maintain the global accuracies of positioning, as well as mapping, and as a matter of course, the aim of this paper is to improve RTK performance in complicated environments by sensor integration. With this purpose, we propose a feature-based RTK/LiDAR/INS integrated system which is designed to utilize the measurements of the repeatedly observed LiDAR features to support RTK, specifically, to improve RTK fix rate. In detail, we adopt tight coupling scheme, after initializing the LiDAR feature map with RTK, we can introduce the measurements of the LiDAR feature with previously obtained global parameters to RTK to estimate a more accurate search center for ambiguities in GNSS-challenged environments, and thus, to improve the success rate of RTK ambiguity resolution. Then, the RTK result derived at each epoch can be used to update the global parameters of the new features, which forms a mode of positioning while mapping. Both theoretical analyses and simulation experiment results prove that integrating with LiDAR feature measurements can improve the performance of RTK in urban environments.