archive
Every paper Pith has read. Search by title, abstract, or pith.
2900 papers in cs.RO · page 14
-
VLMs act as zero-shot ODD sensors for autonomous driving
Operating Within the Operational Design Domain: Zero-Shot Perception with Vision-Language Models
-
Vision models classify driving domains without training data
Operating Within the Operational Design Domain: Zero-Shot Perception with Vision-Language Models
-
Robots compose brick skills to build unseen structures
BrickCraft: Visuomotor Skill Composition with Situated Manual Guidance for Long-Horizon Interlocking Brick Assembly
-
Dynamic state-based memory compilation boosts embodied agents
MemCompiler: Compile, Don't Inject -- State-Conditioned Memory for Embodied Agents
-
State-aware compiler lifts embodied agents up to 129%
MemCompiler: Compile, Don't Inject -- State-Conditioned Memory for Embodied Agents
-
Attention gaps pick failure demos that raise task success rates
How to Utilize Failure Demo Data?: Effective Data Selection for Imitation Learning Using Distribution Differences in Attention Mechanism
-
Attention differences pick failure demos that raise robot success rates
How to Utilize Failure Demo Data?: Effective Data Selection for Imitation Learning Using Distribution Differences in Attention Mechanism
-
Search finds small perturbations that break robot vision 3-7x better
Search-based Robustness Testing of Laptop Refurbishing Robotic Software
-
Action consistency flags reliable futures in robot world models
Is the Future Compatible? Diagnosing Dynamic Consistency in World Action Models
-
Image models turn text commands into robot paths for 160m flights
PathPainter: Transferring the Generalization Ability of Image Generation Models to Embodied Navigation
-
Repeating demos at anchors beats spreading them thin for robot adaptation
Escaping the Diversity Trap in Robotic Manipulation via Anchor-Centric Adaptation
-
Sensor checks let simulated cars use wireless warnings safely
MORPH-U: Multi-Objective Resilient Motion Planning for V2X-Enabled Autonomous Driving in High-Uncertainty Environments via Simulation
-
Radar produces reliable scene captions in fog and snow
Weather-Robust Scene Semantics with Vision-Aligned 4D Radar
-
Cached states cut robot LLM latency 26-fold
CSR: Infinite-Horizon Real-Time Policies with Massive Cached State Representations
-
Adaptive injection adds tactile feedback to VLAs without breaking pretraining
AT-VLA: Adaptive Tactile Injection for Enhanced Feedback Reaction in Vision-Language-Action Models
-
Selective tactile signals improve VLA performance in robot contact tasks
AT-VLA: Adaptive Tactile Injection for Enhanced Feedback Reaction in Vision-Language-Action Models
-
Multi-agent robot system follows biology lab protocols with vision checks
BioProVLA-Agent: An Affordable, Protocol-Driven, Vision-Enhanced VLA-Enabled Embodied Multi-Agent System with Closed-Loop-Capable Reasoning for Biological Laboratory Manipulation
-
Co-contraction tunes aerodynamic damping in dual-rotor actuators at fixed thrust
Variable Aerodynamic Damping via Co-Contraction: A Dynamic Isomorphism with Variable Stiffness Actuators
-
Sparse map guides palm-sized UAVs to explore with low computation
Palm-sized Omnidirectional Vision-Based UAV Exploration with Sparse Topological Map Guidance
-
PISTO raises motion planning success to 89% with smoother paths
PISTO: Proximal Inference for Stochastic Trajectory Optimization
-
Landmark-guided RL drives gastric capsule to 97% coverage
Anatomical Landmark-Guided Deep Reinforcement Learning for Autonomous Gastric Navigation
-
Triangle method recovers camera positions from noisy directions
TriP: A Triangle Puzzle Approach to Robust Translation Averaging
-
Residual feature actions speed up world models for robots
Learning Visual Feature-Based World Models via Residual Latent Action
-
Separable optimization runs radar bundle adjustment on full images
Dr-BA: Separable Optimization for Direct Radar Bundle Adjustment & Localization
-
Context-energy term creates selective forces in Hamiltonian navigation
Learning Material-Aware Hamiltonian Risk Fields for Safe Navigation
-
Estimated target lets robots track accurately at low stiffness
Intention assimilation control for accurate tracking with variable impedance in teleoperation
-
Two drones carry flexible objects by adapting to unknown bend in real time
AirBender: Adaptive Transportation of Bendable Objects Using Dual UAVs
-
Agents can herd on the wrong belief
The Cost of Consensus: Malignant Epistemic Herding and Adaptive Gating in Distributed Multi-Agent Search
-
Language descriptions become solvable constraints for AV tests
Traffic Scenario Orchestration from Language via Constraint Satisfaction
-
CORDIC iteration depth trims 33 percent of inference cycles
CARMEN: CORDIC-Accelerated Resource-Efficient Multi-Precision Inference Engine for Deep Learning
-
Dataset records robot navigation with pairs of humans from two countries
Bi3: A Biplatform, Bicultural, Biperson Dataset for Social Robot Navigation
-
Randomness enables role coordination for identical agents
Randomness is sometimes necessary for coordination
-
Robots coordinate to assist pedestrians crossing roads
Multi-Robot Coordination in V2X Environments
-
Multi-agent RL beats single models on visual-acoustic navigation
Cross-Modal Navigation with Multi-Agent Reinforcement Learning
-
Physics simulation and RL automate motion retargeting for robots
ReActor: Reinforcement Learning for Physics-Aware Motion Retargeting
-
UAV arm positions end effector within centimeters of flowers
An Aerial Manipulator for Perception-Driven Flower Targeting Toward Contactless Pollination in Vertical Farming
-
Invariant decomposition and revision loop reduce drift in 3D layouts
R$^3$L: Reasoning 3D Layouts from Relative Spatial Relations
-
Decomposition and revision reduce errors in 3D layout reasoning
R$^3$L: Reasoning 3D Layouts from Relative Spatial Relations
-
Recursive Lie-group methods compute robot dynamics derivatives quadratically fast
Lie Group Formulation of Recursive Dynamics Algorithms of Higher Order for Floating-Base Robots
-
4B open model matches commercial LLMs on STL translation
ReasonSTL: Bridging Natural Language and Signal Temporal Logic via Tool-Augmented Process-Rewarded Learning
-
4B open model tops natural-language-to-STL translation
ReasonSTL: Bridging Natural Language and Signal Temporal Logic via Tool-Augmented Process-Rewarded Learning
-
Addressable object slots keep robot actions accurate after scene changes
OA-WAM: Object-Addressable World Action Model for Robust Robot Manipulation
-
Dataset pairs ground and drone views for off-road robot teams
GA3T: A Ground-Aerial Terrain Traversability Dataset for Heterogeneous Robot Teams in Unstructured Environments
-
Electronics-free valve turns grasp forces into air-pressure feedback
TouchDrive: Electronics-Free Tactile Sensing Interface for Assistive Grasping
-
1000 hours of human video outperforms 100 hours of robot data
HumanNet: Scaling Human-centric Video Learning to One Million Hours
-
Semantic encoders beat reconstruction for robot world models
Reconstruction or Semantics? What Makes a Latent Space Useful for Robotic World Models
-
Geometry assistance raises novice rope untangling success to 88%
AssistDLO: Assistive Teleoperation for Deformable Linear Object Manipulation
-
VISER benchmark correlates sim and real robot performance at 0.92
Toward Visually Realistic Simulation: A Benchmark for Evaluating Robot Manipulation in Simulation
-
Knowledge transfer boosts robot models to 86% success with 1% parameters
CKT-WAM: Parameter-Efficient Context Knowledge Transfer Between World Action Models
-
Lagrangian GPs preserve energy in learned dynamics models
Structure-Preserving Gaussian Processes Via Discrete Euler-Lagrange Equations