archive
Every paper Pith has read. Search by title, abstract, or pith.
2900 papers in cs.RO · page 11
-
Damped particle motion solves distributed pose graphs
Distributed Pose Graph Optimization via Continuous Riemannian Dynamics
-
RankQ ranks actions to lift offline-to-online RL beyond data
RankQ: Offline-to-Online Reinforcement Learning via Self-Supervised Action Ranking
-
Ranking loss steers Q-learning toward better actions from offline data
RankQ: Offline-to-Online Reinforcement Learning via Self-Supervised Action Ranking
-
Forecasting 3D task outcomes raises pick-and-place success
Forecast-aware Gaussian Splatting for Predictive 3D Representation in Language-Guided Pick-and-Place Manipulation
-
Planner clusters surfaces to shorten UAV inspection flights
ASIP-Planner: Adaptive Planning for UAV Surface Inspection in Partially Known Indoor Environments
-
SEVO transfers VLA policies to new rooms at 75-85% success
SEVO: Semantic-Enhanced Virtual Observation for Robust VLA Manipulation via Active Illumination and Data-Centric Collection
-
Passivity conditions unify fractional haptic rendering
Haptic Rendering of Fractional-Order Viscoelasticity: Passivity and Rendering Fidelity
-
Adaptive model unifies robot prediction and reaction for unseen tasks
HarmoWAM: Harmonizing Generalizable and Precise Manipulation via Adaptive World Action Models
-
Neural tilting of Lévy measures enables jump-preserving SDE inference
Variational Inference for L\'evy Process-Driven SDEs via Neural Tilting
-
Freezing priors in VLA models boosts robot adaptation by 11 points
PriorVLA: Prior-Preserving Adaptation for Vision-Language-Action Models
-
Dual memory buffers and prediction raise robot task success
RoboMemArena: A Comprehensive and Challenging Robotic Memory Benchmark
-
MAPF cast as polynomial linear program via Markovian transport
Optimal and Scalable MAPF via Multi-Marginal Optimal Transport and Schr\"odinger Bridges
-
Multi-agent driving beats single agents in closed-loop benchmark
MDrive: Benchmarking Closed-Loop Cooperative Driving for End-to-End Multi-agent Systems
-
VLA models gain auxiliary skills from parameter-difference vectors
CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models
-
3D magnetic fields let UAVs plan paths 200 times faster than RRT*
Safe Aerial 3D Path Planning for Autonomous UAVs using Magnetic Potential Fields
-
No driving world model excels across all realism measures
Is Your Driving World Model an All-Around Player?
-
Human corrections steer noise to lift robot success from 20% to 90%
Unified Noise Steering for Efficient Human-Guided VLA Adaptation
-
Algebraic latent transitions lift VLA success from 48% to 85% on MetaWorld
ALAM: Algebraically Consistent Latent Action Model for Vision-Language-Action Models
-
Latent model from videos lifts robot success from 48% to 85%
ALAM: Algebraically Consistent Latent Action Model for Vision-Language-Action Models
-
RGB-only multi-agent SLAM matches RGB-D map quality
MAGS-SLAM: Monocular Multi-Agent Gaussian Splatting SLAM for Geometrically and Photometrically Consistent Reconstruction
-
Counterfactual reasoning cuts autonomous driving collisions
C-CoT: Counterfactual Chain-of-Thought with Vision-Language Models for Safe Autonomous Driving
-
Decentralized MPC with safe sets guarantees multi-agent collision avoidance
Decentralized Contingency MPC based on Safe Sets for Nonlinear Multi-agent Collision Avoidance
-
View planner rankings flip with budget and reachability
ObjView-Bench: Rethinking Difficulty and Deployment for Object-Centric View Planning
-
RL xApp cuts UAV handovers by 54% in O-RAN
xApp Empowered Resource Management for Non-Terrestrial Users in 5G O-RAN Networks
-
VRA filters robot accelerations to voltage limits
VRA: Grounding Discrete-Time Joint Acceleration in Voltage-Constrained Actuation
-
VRA restricts accelerations to voltage-realizable bounds
VRA: Grounding Discrete-Time Joint Acceleration in Voltage-Constrained Actuation
-
Embodied AI Success Requires Safe Deployment as Much as Capability
Embodied AI in Action: Insights from SAE World Congress 2026 on Safety, Trust, Robotics, and Real-World Deployment
-
Parallel latent BEV prediction builds long-horizon driving world model
DeepSight: Long-Horizon World Modeling via Latent States Prediction for End-to-End Autonomous Driving
-
ForceFlow lifts robot contact success 37% over baseline
ForceFlow: Learning to Feel and Act via Contact-Driven Flow Matching
-
Early 3D alignment in visual encoders boosts VLA robot performance
VEGA: Visual Encoder Grounding Alignment for Spatially-Aware Vision-Language-Action Models
-
Fused vision-language and geometry align 3D scene graphs
OpenSGA: Efficient 3D Scene Graph Alignment in the Open World
-
RL learns priorities to control communication in agent teams
Priority-Driven Control and Communication in Decentralized Multi-Agent Systems via Reinforcement Learning
-
Model-free RL learns joint control and communication priorities
Priority-Driven Control and Communication in Decentralized Multi-Agent Systems via Reinforcement Learning
-
Inverted culling speeds dynamic LiDAR ray tracing
Geometrically Approximated Modeling for Emitter-Centric Ray-Triangle Filtering in Arbitrarily Dynamic LiDAR Simulation
-
Point clouds modeled as Gaussian statistical manifolds
Learning Point Cloud Geometry as a Statistical Manifold: Theory and Practice
-
VISOR automates robot test oracles using vision-language models
VISOR: A Vision-Language Model-based Test Oracle for Testing Robots
-
VLMs automate robot task oracles from video
VISOR: A Vision-Language Model-based Test Oracle for Testing Robots
-
Driving model accuracy peaks at different frame rates by size
Temporal Sampling Frequency Matters: A Capacity-Aware Study of End-to-End Driving Trajectory Prediction
-
Parts and optical flow improve dynamic Gaussian splatting
PaMoSplat: Part-Aware Motion-Guided Gaussian Splatting for Dynamic Scene Reconstruction
-
Vision Mamba cuts compute for high-res maritime detection
Increasing the Efficiency of DETR for Maritime High-Resolution Images
-
Compact network segments terrain on microcontrollers
Nano-U: Efficient Terrain Segmentation for Tiny Robot Navigation
-
Two-stage framework boosts heterogeneous object manipulation by 31%
HeteroGenManip: Generalizable Manipulation For Heterogeneous Object Interactions
-
Two-stage framework raises manipulation success 31 percent across object types
HeteroGenManip: Generalizable Manipulation For Heterogeneous Object Interactions
-
Transformer 3D affordances raise RL route completion 9% in new towns
MTA-RL: Robust Urban Driving via Multi-modal Transformer-based 3D Affordances and Reinforcement Learning
-
Reranking imagined 3D rollouts raises robot success by 6.8 percent
Data-Asymmetric Latent Imagination and Reranking for 3D Robotic Imitation Learning
-
Physics abstractions raise robot navigation transfer success
Plan in Sandbox, Navigate in Open Worlds: Learning Physics-Grounded Abstracted Experience for Embodied Navigation
-
Frozen VLAs steer generation with memories of their own successes
Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs
-
Frozen VLA adapts by reusing its own successful actions
Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs
-
Cell decomposition adds visibility to simplify 3D path optimization
A cell-decomposition based path planner for 3D navigation in constrained workspaces
-
Assistive forces halve learning time for robot jumps and flips
EFGCL: Learning Dynamic Motion through Spotting-Inspired External Force Guided Curriculum Learning