arXiv subjects

Karthik Desingh

Publications and source records attributed to Karthik Desingh.

At least 19 recordsLinked to original sources

arXiv2026

Active Real-World Factor-Based Evaluation for Generalist Robot Policies

Generalist robot manipulation policies trained on large, diverse datasets have shown remarkable promise across a wide range of tasks. However, rigorously evaluating these policies remains a fundamental challenge. Real-world performance depends on a large combinatorial space of task factors including object poses and camera viewpoints, making full, exhaustive evaluation intractable. Additionally, real hardware evaluation is slow and resource-intensive, so current practice is to use narrow test suites that can miss critical failure modes and misrepresent true deployment readiness. We propose an active evaluation framework that addresses this challenge by treating policy evaluation as a sequential experimental design problem. Our approach fits a probabilistic surrogate model over a structured space of task factors and adaptively selects evaluation configurations to maximize information gain over the policy's performance distribution, allowing for sample-efficient characterization of policy behavior across unseen conditions and a systematic identification of failure-prone regions. We conduct 2331 real-world evaluations across 3 tasks with 3 factor variations and find that our approach typically saves the evaluator at least 20-40% of trials compared to typical random testing.

cs.LG↗

arXiv2026

DREAMSTEER: Latent World Models Can Steer VLA Policies During Deployment Without Any Finetuning

Pretrained vision-language-action (VLA) policies show promising zero-shot generalization, but often fail under deployment-time distribution shift, leading to decreased robustness and inconsistent instruction following. While prior work commonly tackles this by finetuning on in-distribution data, it assumes demonstrations collected on tasks in the target environment. In this work, we propose DREAMSTEER, a deployment-time steering framework for pretrained VLAs without any finetuning or parameter modifications. The key insight in DREAMSTEER is to leverage a latent world model and a value model to steer pretrained VLA policies. During deployment, DREAMSTEER samples candidate action chunks from a VLA policy and predefined motion primitives, imagines their outcomes using an action-conditioned latent world model, and ranks the imagined trajectories with a language-conditioned value model. Across four real-world manipulation benchmarks with unseen objects, DREAMSTEER improves task success rate from 23.75% to 66.25% and instruction-following accuracy from 38.75% to 56.25% over the base VLA policy.

cs.RO↗

arXiv2026

Learning Category-level Last-meter Navigation from RGB Demonstrations of a Single-instance

Achieving precise positioning of the mobile manipulator's base is essential for successful manipulation actions that follow. Most of the RGB-based navigation systems only guarantee coarse, meter-level accuracy, making them less suitable for the precise positioning phase of mobile manipulation. This gap prevents manipulation policies from operating within the distribution of their training demonstrations, resulting in frequent execution failures. We address this gap by introducing an object-centric imitation learning framework for last-meter navigation, enabling a quadruped mobile manipulator robot to achieve manipulation-ready positioning using only RGB observations from its onboard cameras. Our method conditions the navigation policy on three inputs: goal images, multi-view RGB observations from the onboard cameras, and a text prompt specifying the target object. A language-driven segmentation module and a spatial score-matrix decoder then supply explicit object grounding and relative pose reasoning. Using real-world data from a single object instance within a category, the system generalizes to unseen object instances across diverse environments with challenging lighting and background conditions. To comprehensively evaluate this, we introduce two metrics: an edge-alignment metric, which uses ground truth orientation, and an object-alignment metric, which evaluates how well the robot visually faces the target. Under these metrics, our policy achieves 74.58% success in edge-alignment and 89.42% success in object-alignment when positioning relative to unseen target objects. These results show that precise last-meter navigation can be achieved at a category-level without depth, LiDAR, or map priors, enabling a scalable pathway toward unified mobile manipulation. Project page: https://rpm-lab-umn.github.io/category-level-last-meter-nav/

cs.RO↗

arXiv2025

SENT Map -- Semantically Enhanced Topological Maps with Foundation Models

We introduce SENT-Map, a semantically enhanced topological map for representing indoor environments, designed to support autonomous navigation and manipulation by leveraging advancements in foundational models (FMs). Through representing the environment in a JSON text format, we enable semantic information to be added and edited in a format that both humans and FMs understand, while grounding the robot to existing nodes during planning to avoid infeasible states during deployment. Our proposed framework employs a two stage approach, first mapping the environment alongside an operator with a Vision-FM, then using the SENT-Map representation alongside a natural-language query within an FM for planning. Our experimental results show that semantic-enhancement enables even small locally-deployable FMs to successfully plan over indoor environments.

cs.RO↗

arXiv2025

AugInsert: Learning Robust Visual-Force Policies via Data Augmentation for Object Assembly Tasks

Operating in unstructured environments like households requires robotic policies that are robust to out-of-distribution conditions. Although much work has been done in evaluating robustness for visuomotor policies, the robustness evaluation of a multisensory approach that includes force-torque sensing remains largely unexplored. This work introduces a novel, factor-based evaluation framework with the goal of assessing the robustness of multisensory policies in a peg-in-hole assembly task. To this end, we develop a multisensory policy framework utilizing the Perceiver IO architecture to learn the task. We investigate which factors pose the greatest generalization challenges in object assembly and explore a simple multisensory data augmentation technique to enhance out-of-distribution performance. We provide a simulation environment enabling controlled evaluation of these factors. Our results reveal that multisensory variations such as Grasp Pose present the most significant challenges for robustness, and naive unisensory data augmentation applied independently to each sensory modality proves insufficient to overcome them. Additionally, we find force-torque sensing to be the most informative modality for our contact-rich assembly task, with vision being the least informative. Finally, we briefly discuss supporting real-world experimental results. For additional experiments and qualitative results, we refer to the project webpage https://rpm-lab-umn.github.io/auginsert/ .

cs.RO↗

arXiv2025

SPARK-Remote: A Cost-Effective System for Remote Bimanual Robot Teleoperation

Robot teleoperation enables human control over robotic systems in environments where full autonomy is challenging. Recent advancements in low-cost teleoperation devices and VR/AR technologies have expanded accessibility, particularly for bimanual robot manipulators. However, transitioning from in-person to remote teleoperation presents challenges in task performance. We introduce SPARK, a kinematically scaled, low-cost teleoperation system for operating bimanual robots. Its effectiveness is compared to existing technologies like the 3D SpaceMouse and VR/AR controllers. We further extend SPARK to SPARK-Remote, integrating sensor-based force feedback using haptic gloves and a force controller for remote teleoperation. We evaluate SPARK and SPARK-Remote variants on 5 bimanual manipulation tasks which feature operational properties - positional precision, rotational precision, large movements in the workspace, and bimanual collaboration - to test the effective teleoperation modes. Our findings offer insights into improving low-cost teleoperation interfaces for real-world applications. For supplementary materials, additional experiments, and qualitative results, visit the project webpage: https://bit.ly/41EfcJa

cs.RO↗

arXiv2025

SuperQ-GRASP: Superquadrics-based Grasp Pose Estimation on Larger Objects for Mobile-Manipulation

Grasp planning and estimation have been a longstanding research problem in robotics, with two main approaches to find graspable poses on the objects: 1) geometric approach, which relies on 3D models of objects and the gripper to estimate valid grasp poses, and 2) data-driven, learning-based approach, with models trained to identify grasp poses from raw sensor observations. The latter assumes comprehensive geometric coverage during the training phase. However, the data-driven approach is typically biased toward tabletop scenarios and struggle to generalize to out-of-distribution scenarios with larger objects (e.g. chair). Additionally, raw sensor data (e.g. RGB-D data) from a single view of these larger objects is often incomplete and necessitates additional observations. In this paper, we take a geometric approach, leveraging advancements in object modeling (e.g. NeRF) to build an implicit model by taking RGB images from views around the target object. This model enables the extraction of explicit mesh model while also capturing the visual appearance from novel viewpoints that is useful for perception tasks like object detection and pose estimation. We further decompose the NeRF-reconstructed 3D mesh into superquadrics (SQs) -- parametric geometric primitives, each mapped to a set of precomputed grasp poses, allowing grasp composition on the target object based on these primitives. Our proposed pipeline overcomes the problems: a) noisy depth and incomplete view of the object, with a modeling step, and b) generalization to objects of any size. For more qualitative results, refer to the supplementary video and webpage https://bit.ly/3ZrOanU

cs.RO↗

arXiv2024

Talk Through It: End User Directed Manipulation Learning

Training generalist robot agents is an immensely difficult feat due to the requirement to perform a huge range of tasks in many different environments. We propose selectively training robots based on end-user preferences instead. Given a factory model that lets an end user instruct a robot to perform lower-level actions (e.g. 'Move left'), we show that end users can collect demonstrations using language to train their home model for higher-level tasks specific to their needs (e.g. 'Open the top drawer and put the block inside'). We demonstrate this hierarchical robot learning framework on robot manipulation tasks using RLBench environments. Our method results in a 16% improvement in skill success rates compared to a baseline method. In further experiments, we explore the use of the large vision-language model (VLM), Bard, to automatically break down tasks into sequences of lower-level instructions, aiming to bypass end-user involvement. The VLM is unable to break tasks down to our lowest level, but does achieve good results breaking high-level tasks into mid-level skills. We have a supplemental video and additional results at talk-through-it.github.io.

cs.RO↗

arXiv2024

Evaluating Robustness of Visual Representations for Object Assembly Task Requiring Spatio-Geometrical Reasoning

This paper primarily focuses on evaluating and benchmarking the robustness of visual representations in the context of object assembly tasks. Specifically, it investigates the alignment and insertion of objects with geometrical extrusions and intrusions, commonly referred to as a peg-in-hole task. The accuracy required to detect and orient the peg and the hole geometry in SE(3) space for successful assembly poses significant challenges. Addressing this, we employ a general framework in visuomotor policy learning that utilizes visual pretraining models as vision encoders. Our study investigates the robustness of this framework when applied to a dual-arm manipulation setup, specifically to the grasp variations. Our quantitative analysis shows that existing pretrained models fail to capture the essential visual features necessary for this task. However, a visual encoder trained from scratch consistently outperforms the frozen pretrained models. Moreover, we discuss rotation representations and associated loss functions that substantially improve policy learning. We present a novel task scenario designed to evaluate the progress in visuomotor policy learning, with a specific focus on improving the robustness of intricate assembly tasks that require both geometrical and spatial reasoning. Videos, additional experiments, dataset, and code are available at https://bit.ly/geometric-peg-in-hole .

cs.RO↗

arXiv2023

SlotGNN: Unsupervised Discovery of Multi-Object Representations and Visual Dynamics

Learning multi-object dynamics from visual data using unsupervised techniques is challenging due to the need for robust, object representations that can be learned through robot interactions. This paper presents a novel framework with two new architectures: SlotTransport for discovering object representations from RGB images and SlotGNN for predicting their collective dynamics from RGB images and robot interactions. Our SlotTransport architecture is based on slot attention for unsupervised object discovery and uses a feature transport mechanism to maintain temporal alignment in object-centric representations. This enables the discovery of slots that consistently reflect the composition of multi-object scenes. These slots robustly bind to distinct objects, even under heavy occlusion or absence. Our SlotGNN, a novel unsupervised graph-based dynamics model, predicts the future state of multi-object scenes. SlotGNN learns a graph representation of the scene using the discovered slots from SlotTransport and performs relational and spatial reasoning to predict the future appearance of each slot conditioned on robot actions. We demonstrate the effectiveness of SlotTransport in learning object-centric features that accurately encode both visual and positional information. Further, we highlight the accuracy of SlotGNN in downstream robotic tasks, including challenging multi-object rearrangement and long-horizon prediction. Finally, our unsupervised approach proves effective in the real world. With only minimal additional data, our framework robustly predicts slots and their corresponding dynamics in real-world control tasks.

cs.RO↗

arXiv2023

DNBP: Differentiable Nonparametric Belief Propagation

We present a differentiable approach to learn the probabilistic factors used for inference by a nonparametric belief propagation algorithm. Existing nonparametric belief propagation methods rely on domain-specific features encoded in the probabilistic factors of a graphical model. In this work, we replace each crafted factor with a differentiable neural network enabling the factors to be learned using an efficient optimization routine from labeled data. By combining differentiable neural networks with an efficient belief propagation algorithm, our method learns to maintain a set of marginal posterior samples using end-to-end training. We evaluate our differentiable nonparametric belief propagation (DNBP) method on a set of articulated pose tracking tasks and compare performance with learned baselines. Results from these experiments demonstrate the effectiveness of using learned factors for tracking and suggest the practical advantage over hand-crafted approaches. The project webpage is available at: https://progress.eecs.umich.edu/projects/dnbp/ .

cs.RO↗

arXiv2022

SORNet: Spatial Object-Centric Representations for Sequential Manipulation

Sequential manipulation tasks require a robot to perceive the state of an environment and plan a sequence of actions leading to a desired goal state. In such tasks, the ability to reason about spatial relations among object entities from raw sensor inputs is crucial in order to determine when a task has been completed and which actions can be executed. In this work, we propose SORNet (Spatial Object-Centric Representation Network), a framework for learning object-centric representations from RGB images conditioned on a set of object queries, represented as image patches called canonical object views. With only a single canonical view per object and no annotation, SORNet generalizes zero-shot to object entities whose shape and texture are both unseen during training. We evaluate SORNet on various spatial reasoning tasks such as spatial relation classification and relative direction regression in complex tabletop manipulation scenarios and show that SORNet significantly outperforms baselines including state-of-the-art representation learning techniques. We also demonstrate the application of the representation learned by SORNet on visual-servoing and task planning for sequential manipulation on a real robot.

cs.RO↗

arXiv2022

Break and Make: Interactive Structural Understanding Using LEGO Bricks

Visual understanding of geometric structures with complex spatial relationships is a fundamental component of human intelligence. As children, we learn how to reason about structure not only from observation, but also by interacting with the world around us -- by taking things apart and putting them back together again. The ability to reason about structure and compositionality allows us to not only build things, but also understand and reverse-engineer complex systems. In order to advance research in interactive reasoning for part-based geometric understanding, we propose a challenging new assembly problem using LEGO bricks that we call Break and Make. In this problem an agent is given a LEGO model and attempts to understand its structure by interactively inspecting and disassembling it. After this inspection period, the agent must then prove its understanding by rebuilding the model from scratch using low-level action primitives. In order to facilitate research on this problem we have built LTRON, a fully interactive 3D simulator that allows learning agents to assemble, disassemble and manipulate LEGO models. We pair this simulator with a new dataset of fan-made LEGO creations that have been uploaded to the internet in order to provide complex scenes containing over a thousand unique brick shapes. We take a first step towards solving this problem using sequence-to-sequence models that provide guidance for how to make progress on this challenging problem. Our simulator and data are available at github.com/aaronwalsman/ltron. Additional training code and PyTorch examples are available at github.com/aaronwalsman/ltron-torch-eccv22.

cs.CV↗

arXiv2021

LanguageRefer: Spatial-Language Model for 3D Visual Grounding

For robots to understand human instructions and perform meaningful tasks in the near future, it is important to develop learned models that comprehend referential language to identify common objects in real-world 3D scenes. In this paper, we introduce a spatial-language model for a 3D visual grounding problem. Specifically, given a reconstructed 3D scene in the form of point clouds with 3D bounding boxes of potential object candidates, and a language utterance referring to a target object in the scene, our model successfully identifies the target object from a set of potential candidates. Specifically, LanguageRefer uses a transformer-based architecture that combines spatial embedding from bounding boxes with fine-tuned language embeddings from DistilBert to predict the target object. We show that it performs competitively on visio-linguistic datasets proposed by ReferIt3D. Further, we analyze its spatial reasoning task performance decoupled from perception noise, the accuracy of view-dependent utterances, and viewpoint annotations for potential robotics applications.

cs.RO↗

arXiv2021

Differentiable Nonparametric Belief Propagation

We present a differentiable approach to learn the probabilistic factors used for inference by a nonparametric belief propagation algorithm. Existing nonparametric belief propagation methods rely on domain-specific features encoded in the probabilistic factors of a graphical model. In this work, we replace each crafted factor with a differentiable neural network enabling the factors to be learned using an efficient optimization routine from labeled data. By combining differentiable neural networks with an efficient belief propagation algorithm, our method learns to maintain a set of marginal posterior samples using end-to-end training. We evaluate our differentiable nonparametric belief propagation (DNBP) method on a set of articulated pose tracking tasks and compare performance with a recurrent neural network. Results from this comparison demonstrate the effectiveness of using learned factors for tracking and suggest the practical advantage over hand-crafted approaches. The project webpage is available at: progress.eecs.umich.edu/projects/dnbp.

cs.RO↗

arXiv2020

Parts-Based Articulated Object Localization in Clutter Using Belief Propagation

Robots working in human environments must be able to perceive and act on challenging objects with articulations, such as a pile of tools. Articulated objects increase the dimensionality of the pose estimation problem, and partial observations under clutter create additional challenges. To address this problem, we present a generative-discriminative parts-based recognition and localization method for articulated objects in clutter. We formulate the problem of articulated object pose estimation as a Markov Random Field (MRF). Hidden nodes in this MRF express the pose of the object parts, and edges express the articulation constraints between parts. Localization is performed within the MRF using an efficient belief propagation method. The method is informed by both part segmentation heatmaps over the observation, generated by a neural network, and the articulation constraints between object parts. Our generative-discriminative approach allows the proposed method to function in cluttered environments by inferring the pose of occluded parts using hypotheses from the visible parts. We demonstrate the efficacy of our methods in a tabletop environment for recognizing and localizing hand tools in uncluttered and cluttered configurations.

cs.RO↗

arXiv2020

A Sketch-Based System for Human-Guided Constrained Object Manipulation

In this paper, we present an easy to use sketch-based interface to extract geometries and generate affordance files from 3D point clouds for robot-object interaction tasks. Using our system, even novice users can perform robot task planning by employing such sketch tools. Our focus in this paper is employing human-in-the-loop approach to assist in the generation of more accurate affordance templates and guidance of robot through the task execution process. Since we do not employ any unsupervised learning to generate affordance templates, our system performs much faster and is more versatile for template generation. Our system is based on the extraction of geometries for generalized cylindrical and cuboid shapes, after extracting the geometries, affordances are generated for objects by applying simple sketches. We evaluated our technique by asking users to define affordances by employing sketches on the 3D scenes of a door handle and a drawer handle and used the resulting extracted affordance template files to perform the tasks of turning a door handle and opening a drawer by the robot.

cs.HC↗

arXiv2018

Factored Pose Estimation of Articulated Objects using Efficient Nonparametric Belief Propagation

Robots working in human environments often encounter a wide range of articulated objects, such as tools, cabinets, and other jointed objects. Such articulated objects can take an infinite number of possible poses, as a point in a potentially high-dimensional continuous space. A robot must perceive this continuous pose to manipulate the object to a desired pose. This problem of perception and manipulation of articulated objects remains a challenge due to its high dimensionality and multi-modal uncertainty. In this paper, we propose a factored approach to estimate the poses of articulated objects using an efficient nonparametric belief propagation algorithm. We consider inputs as geometrical models with articulation constraints, and observed RGBD sensor data. The proposed framework produces object-part pose beliefs iteratively. The problem is formulated as a pairwise Markov Random Field (MRF) where each hidden node (continuous pose variable) is an observed object-part's pose and the edges denote the articulation constraints between the parts. We propose articulated pose estimation by Pull Message Passing algorithm for Nonparametric Belief Propagation (PMPNBP) and evaluate its convergence properties over scenes with articulated objects.

cs.RO↗