IPAB Workshop - 13/03/2025 Speaker: Jianning Deng Title: Articulate your NeRF: Unsupervised articulated object modeling via conditional view synthesis Abstract: We propose a novel unsupervised method to learn pose and part-segmentation of articulated objects with rigid parts. Given two observations of an object in different articulation states, our method learns the geometry and appearance of object parts by fitting an implicit model on the first observation and renders the latter observation by distilling the part segmentation and articulation. Additionally, to tackle the challenging joint optimization of part segmentation and articulation, we propose a voxel grid based initialization strategy and a decoupled optimization procedure. Compared to the prior unsupervised work, our model obtains significantly better performance, generalizes to objects with arbitrary number of parts while it can be efficiently learned from few views only for the latter observation. Speaker: Aditya Kamireddypalli Title: ContactFusion: Stochastic Poisson Surface Mapsfrom Visual and Contact Sensing Abstract: "Robust and precise robotic assembly entails inser-tion of constituent components. Insertion success is hinderedwhen noise in scene understanding exceeds tolerance limits,especially when fabricated with tight tolerances. In this work,we propose ContactFusion which combines global mappingwith local contact information, fusing point clouds with forcesensing. Our method entails a Rejection Sampling based contactoccupancy sensing procedure which estimates contact locationson the end-effector from Force/Torque sensing at the wrist.We demonstrate how to fuse contact with visual informationinto a Stochastic Poisson Surface Map (SPSMap) - a maprepresentation that can be updated with the Stochastic PoissonSurface Reconstruction (SPSR) algorithm. We first validate thecontact occupancy sensor in simulation and show its abilityto detect the contact location on the robot from force sensinginformation. Then, we evaluate our method in a peg-in-holetask, demonstrating an improvement in the hole pose estimatewith the fusion of the contact information with the SPSMap" Mar 13 2025 13.00 - 14.00 IPAB Workshop - 13/03/2025 Jianning Deng & Aditya Kamireddypalli G.03
IPAB Workshop - 13/03/2025 Speaker: Jianning Deng Title: Articulate your NeRF: Unsupervised articulated object modeling via conditional view synthesis Abstract: We propose a novel unsupervised method to learn pose and part-segmentation of articulated objects with rigid parts. Given two observations of an object in different articulation states, our method learns the geometry and appearance of object parts by fitting an implicit model on the first observation and renders the latter observation by distilling the part segmentation and articulation. Additionally, to tackle the challenging joint optimization of part segmentation and articulation, we propose a voxel grid based initialization strategy and a decoupled optimization procedure. Compared to the prior unsupervised work, our model obtains significantly better performance, generalizes to objects with arbitrary number of parts while it can be efficiently learned from few views only for the latter observation. Speaker: Aditya Kamireddypalli Title: ContactFusion: Stochastic Poisson Surface Mapsfrom Visual and Contact Sensing Abstract: "Robust and precise robotic assembly entails inser-tion of constituent components. Insertion success is hinderedwhen noise in scene understanding exceeds tolerance limits,especially when fabricated with tight tolerances. In this work,we propose ContactFusion which combines global mappingwith local contact information, fusing point clouds with forcesensing. Our method entails a Rejection Sampling based contactoccupancy sensing procedure which estimates contact locationson the end-effector from Force/Torque sensing at the wrist.We demonstrate how to fuse contact with visual informationinto a Stochastic Poisson Surface Map (SPSMap) - a maprepresentation that can be updated with the Stochastic PoissonSurface Reconstruction (SPSR) algorithm. We first validate thecontact occupancy sensor in simulation and show its abilityto detect the contact location on the robot from force sensinginformation. Then, we evaluate our method in a peg-in-holetask, demonstrating an improvement in the hole pose estimatewith the fusion of the contact information with the SPSMap" Mar 13 2025 13.00 - 14.00 IPAB Workshop - 13/03/2025 Jianning Deng & Aditya Kamireddypalli G.03
