In this research, we propose a novel wheel-legged mobile robot to address the problems of insufficient obstacle-crossing performance and poor motion flexibility of mobile robots in non-structural environments. Firstly, we designed the transformable wheel mechanism and tail adaptive mechanism. Secondly, the kinematic model of the robot is established and solved by analyzing the whole motion and wheel-legged switching motion for the operation requirements under different road conditions. By synthesizing the constraint relationships among the modules and analyzing the robot’s obstacle-crossing abilities, we systematically established the mechanical model of the robot when it encounters obstacles. Thirdly, we studied the stability of the robot based on the stable cone method in the case of slope and unilateral transformation wheel deployment and achieved the tipping condition in the critical state. Finally, we used ADAMS software to simulate and analyze the driving process of the robot in various types of terrain and obstacles in order to verify that it has superior performance through obstacles and motion flexibility. The analysis shows that the robot can passively adapt to various complex and variable obstacle-filled terrains with obstacle heights which are much higher than its center of gravity range. The results of the study can provide a reference for the structural optimization and the obstacle-crossing performance improvement of mobile robots.

Simultaneous localization and mapping (SLAM) is fundamental for intelligent mobile units to perform diverse tasks. Recent work has shown that integrating neural rendering and SLAM showed promising results in photorealistic environment reconstruction. However, existing methods estimate pose by minimizing the error between rendered and input images, which is time-consuming and cannot be run in real-time, deviating from the original intention of SLAM. In this paper, we propose a dense RGB-D SLAM based on 3D Gaussian splatting (3DGS) while employing generalized iterative closest point (G-ICP) for pose estimation. We actively utilize 3D point cloud information to improve the tracking accuracy and operating speed of the system. At the same time, we propose a dual keyframe selection strategy and its corresponding densification method, which can effectively reconstruct new observation scenes and improve the quality of previously constructed maps. In addition, we introduce regularization loss to address the scale explosion of the 3D Gaussians and over-elongate in the camera viewing direction. Experiments on the Replica, TUM-RGBD, and ScanNet datasets show that our method achieves state-of-the-art tracking accuracy and runtime while being competitive in rendering quality.