IPAB Workshop-20/06/2024

Speaker: Miaowei Wang

Title: Canonical Consolidation Fields: Reconstructing Dynamic Shapes from Point Clouds

Abstract: We present Canonical Consolidation Fields (CanFields): a method for reconstructing a time series of independently-sampled point clouds into a single deforming coherent shape. Such input often comes from motion capture. Existing methods either couple the geometry and the deformation, where by doing so they smooth fine details and lose the ability to track moving points, or they track the deformation explicitly, but introduce topological and geometric artifacts. Our novelty lies in the consolidation of the point clouds into a single canonical shape in a way that reduces the effect of noise and outliers and enables us to overcome missing regions. We simultaneously reconstruct the velocity fields that guide the deformation. This consolidation allows us to retain the high-frequency details of the geometry, while faithfully reproducing the low-frequency deformation. Our architecture comprises simple components, and fits any single input shape without using datasets. We demonstrate the robustness and accuracy of our methods on a diverse benchmark of dynamic point clouds, including missing regions, sparse frames, and noise.

Speaker: Victor Leve Usage Leve-Lin

Title: Explicit Contact and Motion Planning of Whole-Body Contact-Rich Manipulation

Abstract: The Whole-Body Contact-Rich Manipulation - manipulating objects using contact possibly on the whole robot surface - is one of the most challenging and under-explored research area for modern humanoid robots. Allowing contact location to be on the whole surface of the robot adds a dimension that the state-of-the-art methods can hardly scale to. This work proposes an explicit representation for contact location on the robot surface that allows the formulation of Continuous Optimization problems for Contact and Motion Planning of Whole-Body Contact-Rich Manipulation. Results with this approach on planar scenario showed a significant improvement in convergence to a feasible solution. Furthermore, the proposed representation of the contact location provides a foundational formulation for future research on Whole-Body Contact-Rich Manipulation using Continuous Optimization.