archive
Every paper Pith has read. Search by title, abstract, or pith.
2900 papers in cs.RO · page 7
-
Outcome-variance masking cuts VLA RL gradients to under 20%
Learn Where Outcomes Diverge: Efficient VLA RL via Probabilistic Chunk Masking
-
Dual system turns task instructions into stable tabletop scenes
STABLE: Simulation-Ready Tabletop Layout Generation via a Semantics-Physics Dual System
-
Robots hiding in niches preferred for yielding in evacuations
Beyond Collision Avoidance: Multi-Robot Yielding and Spatial Affordance in Emergency Evacuations
-
Attention weights from a driving detector give real-time explanations
Towards Trustworthy and Explainable AI for Perception Models: From Concept to Prototype Vehicle Deployment
-
Trustworthy perception module runs in prototype vehicle
Towards Trustworthy and Explainable AI for Perception Models: From Concept to Prototype Vehicle Deployment
-
Health projector lets VLA robots handle degraded joints
Health-Conditioned Vision-Language-Action Models for Malfunction-Aware Robot Control
-
Sim states cut rope policy error by 30.8% over vision
Learning Sim-Grounded Policies for Bimanual Rope Manipulation from Human Teleoperation Data
-
Adversarial latent jumps speed up learning 1.67 times
Mind Dreamer: Untethering Imagination via Active Causal Intervention on Latent Manifolds
-
RL predictor lets quadrotors track paths despite mass shifts and slung loads
Adaptive Outer-Loop Control of Quadrotors via Reinforcement Learning
-
Expanding circular regions from LiDAR speed robot local planning
Fast Expanding Safe Circular Regions for Efficient Local Path Planning
-
Exponential cost lets quadrotors pass through narrow gaps
Constrained MPC-Based Motion Planning for Morphing Quadrotors in Ultra-Narrow Passages under Limited Perception
-
Bilevel policies solve 10,000-object long-horizon tasks
Learning Bilevel Policies over Symbolic World Models for Long-Horizon Planning
-
Human preferences guide robot RL with less intervention
OHP-RL: Online Human Preference as Guidance in Reinforcement Learning for Robot Manipulation
-
Autoregressive model steers drones by predicting next world states
WorldVLN: Autoregressive World Action Model for Aerial Vision-Language Navigation
-
300 ms latency with high bitrate matches best-case speed
Driving Through the Network: Performance and Workload Under Latency and Video Impairment
-
Pipeline turns low-cost arm dynamics into feasible models
A Reproducible and Physically Feasible Dynamic Parameter Identification Framework for a Low-Cost Robot Arm
-
Frequency regularization across chunks yields coherent robot actions
FocalPolicy: Frequency-Optimized Chunking and Locally Anchored Flow Matching for Coherent Visuomotor Policy
-
Frequency regularization reduces robot action jumps between chunks
FocalPolicy: Frequency-Optimized Chunking and Locally Anchored Flow Matching for Coherent Visuomotor Policy
-
RL agent tracks tokamak plasma shapes with sensor failures
Dynamic Plasma Shape Control with Arbitrary Sensor Subsets
-
Design guidelines address robot wranglers' individual and team needs
Designing for Robot Wranglers: A Synthesis of Literature and Practice
-
Structured Jacobians speed up high-order motion optimization
Structured Jacobian Construction for Motion Optimization with High-Order Time Derivatives in Multi-Link Systems
-
Geometric pre-training raises robot success with scarce demos
GAP: Geometric Anchor Pre-training for Data-Efficient Visuomotor Learning of Manipulation Tasks
-
PIBT scales better than PP or LNS2 for large AGV teams
From Gridworlds to Warehouses: Adapting Lightweight One-shot Multi-Agent Pathfinding for AGVs
-
LIDAR points yield time-varying CBF safety constraints for robots
Reactive Robot-Centric Safety for Autonomous Navigation in Constrained and Dynamic Environments
-
Geometric method hands off vehicle tracks between UAVs
A Topology-Aware Spatiotemporal Handover Framework for Continuous Multi-UAV Tracking
-
Lamarckian only underperforms in conflicting unpredictable settings
Lamarckian Inheritance in Dynamic Environments: How Key Variables Affect Evolutionary Dynamics
-
Two-drone line-segment inspection is strongly NP-hard
Optimizing Line Segment Inspection with Limited-Range Drones
-
Method infers detailed functional 3D scene graphs in cluttered rooms
Hierarchical and Holistic Open-Vocabulary Functional 3D Scene Graphs for Indoor Spaces
-
Disentangled actions boost video model quality and abstraction
DiLA: Disentangled Latent Action World Models
-
Mass-aware RL lets quadruped arms pick and place while walking
Learning Dynamic Pick-and-Place for a Legged Manipulator
-
Online feedback cuts robot prediction errors by up to 76%
Feedback World Model Enables Precise Guidance of Diffusion Policy
-
Closed-loop testing raises traffic scenario accuracy by 12%
PCASim: Promptable Closed-loop Adversarial Simulation for Urban Traffic Environment
-
New benchmark tests AI on realistic sports moves
MyoChallenge 2025: A New Benchmark for Human Athletic Intelligence
-
One compromised robot spreads unsafe actions to entire team
Propagating Unsafe Actions in LLM Controlled Multi-Robot Collaboration via Single Robot Compromise
-
Planner keeps fixed-wing glider on efficient path despite wind
Wind-Aware Optimal Trajectory Planning for Efficient Gliding of Fixed-Wing Aerial Systems
-
Greedy proxy selects 25 panoramic views per scene while preserving coverage
CM-EVS: Sparse Panoramic RGB-D-Pose Data for Complete Scene Coverage
-
Fine-tuning for noise and latency enables zero-shot RL robot navigation
NavRL++: A System-Level Framework for Improving Sim-to-Real Transfer in Reinforcement Learning-Based Robot Navigation
-
KaRMA ranks robot hands by reachable rolling motions
KaRMA: A Kinematic Metric for Fine Manipulation Ability in Robotic Hands
-
Single policy skips 15-40% of robot steps without losing success
SkiP: When to Skip and When to Refine for Efficient Robot Manipulation
-
Open platform benchmarks MARL for AUV swarms in underwater tracking
Task-Semantic Graph-Driven Distributed Agent Networking for Underwater Target Tracking
-
Terrain-consistent references improve humanoid RL tracking on rough ground
Terrain Consistent Reference-Guided RL for Humanoid Navigation Autonomy
-
Hand robot designs encoded as quadratic optimization problems
A QUBO Formulation Framework for Kinematic Structure-Based Robot Design Optimization: A Robotic Hand Case Study
-
Framework unifies safety filters and auditability for parallel RL
parallelcbf: A composable safety-filter and auditability framework for tensor-parallel reinforcement learning
-
Two-stage model cuts vessel trajectory errors by 25 percent over 10 hours
Hierarchical Two-Stage Framework for Environment-Aware Long-Horizon Vessel Trajectory Prediction
-
Active replay raises LiDAR map recall by 4.66 points
LAPS: Improving Incremental LiDAR Mapping using Active Pooling and Sampling for Neural Distance Fields
-
Legendre polynomials let robot AI act in 31 milliseconds
FLASH: Efficient Visuomotor Policy via Sparse Sampling
-
LLM agents optimize robot task schedules for construction
Hybrid LLM-based Intelligent Framework for Robot Task Scheduling
-
LSTM plus residual RL keeps robot teleoperation stable under delays
Residual Reinforcement Learning for Robot Teleoperation under Stochastic Delays
-
Vision method selects perches on 76% of urban trees
Where to Perch in a Tree: Vision-Guidance for Tree-Grasping Drones
-
Mixed reality merges three robots' maps at 8.83 Hz
MR-SLAM: Immersive Spatial Supervision for Multi-Robot Mapping via Mixed Reality