Due to the presence of ambiguities caused by sensor noise and structural similarity, simultaneous localization and mapping (SLAM) observation models are typically multimodal. The multimodal inference process can be directly dealt with by belief propagation (BP) using weighted Gaussian mixture messages, but for efficiency, a combinatorial explosion of the complexity must be suitably relaxed. In this study, we present an effective multimodal BP SLAM for robust inference with ambiguities. Using Gaussian bandwidth mean shift and cluster-based reparameterization, we reduce the number of Gaussian components in each message due to the BP nature. The proposed algorithm reduces the number of components of the product by summarizing indistinguishable modes in weighted Gaussian mixtures and keeping only the significant modes, making BP computationally efficient.

Welding is very difficult and dangerous for manual welders. The double hull structures found in ships are very hazardous environments, and as such the shipbuilding industry demands a safer, autonomous system to perform the welding rather than deploy manual welders. This paper describes the design of a new mechanism, called the 'Rail Runner', which is able to autonomously travel within the double hull structure. The design of a 3P3R serial manipulator for welding is also described in this paper. As an application of the 'Rail Runner' mechanism, we combine the 'Rail Runner' platform and the 3P3R serial manipulator for autonomous welding. The mechanical system of this robot is composed of a six-axes (3P3R) manipulator for achieving the welding function and a six-axes mobile platform for traveling within the double hull structure. This robot is able to autonomously travel between longitudinal structures, with transverse direction, and is capable of welding in double hull structures. The 'Rail Runner' can raise the efficiency of the welding process, as compared to manual welders. This in turn raises the international competitiveness of the shipbuilding industry.

Given a polygonal mesh with a set of tagged control polygons, a subdivision surface that interpolates the B-spline curves of these control polygons can be obtained by constructing a set of polygonal complexes, each of which converges to one of the given curves. The construction process will geometrically and topologically disturb the initial and/or the subdivided polygonal meshes in the vicinity of the polygonal complexes. This may result in poorly shaped surfaces across the interpolated curves. Based on signal fairing, this paper describes a method to fair such a surface and paves the way to achieve further constraints such as interpolating curves with predefined tangent plane or cross curvature.