is an open access repository that collects the work of Arts et Métiers ParisTech researchers and makes it freely available over the web where possible. a b s t r a c tThe direct metal deposition (DMD) with laser is a free-form metal deposition process for manufacturing dense pieces, which allows generating a prototype or small series of near net-shape structures. One of the most critical issues is that produced pieces have a deleterious surface finish which systematically requires post machining steps. This problem has never been fully addressed before. The present work describes investigations on the DMD process, using an Yb-YAG disk laser, and a widely used titanium alloy (Ti-6Al-4V) to understand the influence of the main process parameters on the surface finish quality. The focus of our work was: (1) to understand the physical mechanisms responsible for deleterious surface finishes, (2) to propose different experimental solutions for improving surface finish.In order to understand the physical mechanisms responsible for deleterious surface finishes, we have carried out: (1) a precise characterization of the laser beam and the powder stream; (2) a large number of multi-layered walls using different process parameters (P(W), V(m/min), D m (g/min), Gaussian or uniform beam distribution); (3) a real time fast camera analysis of melt pool dynamics and melt-pool -powder stream coupling; (4) a characterization of wall morphologies versus process parameters using 2D and 3D profilometry.The results confirm that surface degradation depends on two distinct aspects: the sticking of nonmelted or partially melted particles on the free surfaces, and the formation of menisci with more or less pronounced curvature radii. Among other aspects, a reduction of layer thickness and an increase of melt-pool volumes to favor re-melting processes are shown to have a beneficial effect on roughness parameters. Last, a simple analytical model was proposed to correlate melt-pool geometries to resulting surface finishes.

This paper presents a new method for planning motions of multi-arm systems in constrained workspaces, for which state-of-the-art planners behave poorly. The method is based on the decomposition of the system into parts. Compact roadmaps are first computed for each part, and then, a super-graph is constructed by the composition of elementary roadmaps. Results presented for a three-arm system and a model of the complex DLR's Justin robot show a significant performance gain of such a two-stage roadmap construction method with respect to single-stage methods applied to the whole system.