Tagged articles
imitation learning
20 articles · Page 1 of 1
DexRoam aligns human whole-body demonstrations to robot actions, enabling mobile bimanual dexterous manipulation learning from egocentric human videos and cutting robot teleoperation data needs by half.
CoRL 2026DexRoambimanual manipulation
0 likes · 11 min read

Researchers from HKUST, Beijing Academy of AI, and Peking University present DexRoam, a system that aligns human whole-body demonstrations to robot actions via a three-stage pipeline, enabling mobile bimanual dexterous manipulation learning from egocentric human data while cutting robot demonstration needs by roughly 50%.
CoRL 2026DexRoamRobotics
0 likes · 11 min read

This 72-page survey explores how optimal transport provides geometric tools for comparing distributions in reinforcement learning, covering imitation learning, policy optimization, model-based RL, offline RL, robustness, and transfer, while analyzing design choices in cost functions, temporal handling, and computational trade-offs.
Distribution MatchingOffline RLOptimal Transport
0 likes · 35 min read

LeaP introduces a learnable source prior conditioned on proprioception that jointly predicts mean and variance of a diagonal Gaussian for initializing generative robot policies, achieving 81.6% average success on 15 RoboTwin tasks (25.5 pp over standard Gaussian) and 80% on real Franka tasks (33.3 pp gain), with faster convergence.
CoRL 2026Flow MatchingGenerative Robot Policies
0 likes · 11 min read

Berkeley PhD thesis presents a four-part framework for generalist robots: BridgeData V2 dataset (60k trajectories), Octo policy (800k demos, 29% better zero-shot), CrossFormer (single policy across 20 embodiments, 73% success), and MEM (multi-scale memory for 15-min tasks).
BridgeData V2Cross-EmbodimentCrossFormer
0 likes · 19 min read

Action Map Policy (AMP) reframes robot manipulation as pixel-level classification by projecting 3D keypoint trajectories onto 2D image planes, enabling cross-entropy loss to model multi-modal action distributions with sub-millimeter precision and single-forward-pass inference.
Action Map PolicyMulti-View TransformerPixel Classification
0 likes · 12 min read

The RL‑100 framework combines imitation learning from human tele‑operation with offline and online reinforcement learning to refine diffusion‑based control policies, achieving 100 % success across eight real‑world robot tasks, matching or surpassing human operators in speed while maintaining stability and low latency.
RL-100Reinforcement LearningRobotics
0 likes · 7 min read

The article analyzes the fundamental bottleneck of current Physical AI post‑training, where VLA models lack foresight, and presents World Action Models (WAM) – both Cascaded and Joint architectures – along with four post‑training pathways that combine evaluation and imagination to break the data‑collection loop.
Physical AIReinforcement LearningSimulation
0 likes · 18 min read

In a detailed interview, Ted Xiao, former Google DeepMind researcher, walks through the existence‑proof, foundation‑model, and scaling eras of embodied robot learning, explaining the technical challenges, pivotal decisions, and the evolving role of large language and vision models in robotics.
Embodied AIReinforcement Learningfoundation models
0 likes · 19 min read

This survey maps the 2021‑2025 progress of imitation learning for dexterous manipulation, detailing theoretical foundations, datasets, algorithms, hardware platforms, and evaluation protocols, and highlights challenges such as data quality, hardware dependence, and the need for standardized benchmarks to advance embodied AI.
AlgorithmsRoboticsdatasets
0 likes · 11 min read

The article analyzes the simulation‑reality gap in algorithmic trading and proposes an interactive market simulator that combines a pool of imitation‑learning agents, an action‑synthesis network, and a DDPG‑based reinforcement‑learning trader, showing superior robustness and downside protection on QQQ data.
DDPGReinforcement Learningagent-based modeling
0 likes · 16 min read

This guide demonstrates an end‑to‑end workflow on Alibaba Cloud PAI that uses the Cosmos world model to replace Isaac simulation for robot action data augmentation, including minimal human demonstrations, prompt‑driven data expansion, rejection sampling, IDM inverse‑kinematics extraction, imitation‑learning fine‑tuning, and model evaluation.
AICosmosData Augmentation
0 likes · 17 min read

This article walks through an end‑to‑end Physical AI workflow on Alibaba Cloud PAI, covering robot teleoperation data collection, Isaac Lab‑based data augmentation and enhancement, imitation‑learning model training, distributed DLC execution, and systematic evaluation across varied visual conditions.
Data AugmentationPhysical AIRobotics
0 likes · 17 min read


Amap TechOct 5, 2025 · Artificial Intelligence
CE-Nav introduces a two‑stage imitation‑then‑reinforcement framework that decouples generic geometric planning from robot‑specific dynamics, enabling low‑cost, high‑performance navigation across quadrupeds, humanoids, and drones while requiring only brief online fine‑tuning.
Cross-EmbodimentReinforcement LearningRobot Navigation
0 likes · 11 min read

This article provides a comprehensive, step‑by‑step exploration of MaxEnt Inverse Reinforcement Learning, covering its statistical foundations, feature‑expectation matching, algorithmic details, deep extensions, and practical engineering considerations for complex decision‑making tasks.
Deep IRLFeature Matchingimitation learning
0 likes · 21 min read

The article analyzes the recent surge in embodied intelligence, examines why physical agents matter despite advances in large language models, outlines common failure modes, discusses key research decisions such as 2D versus 3D perception and tactile sensing, and explores the roles of imitation learning, VLA, and reinforcement learning in shaping the field.
Reinforcement LearningRoboticsVLA
0 likes · 24 min read

Didi balances ride‑hailing supply and demand by forecasting regional needs with time‑series and deep‑learning models, then optimally repositioning drivers through integer programming and refining policies via imitation and offline reinforcement learning, ultimately enhancing passenger experience and platform efficiency.
DidiOffline Reinforcement Learningforecasting
0 likes · 16 min read

This article examines the technical challenges of intelligent replenishment—model stability, complexity, generalization, and interpretability—and explains how a few‑shot imitation learning and inverse reinforcement learning framework can overcome these issues to deliver reliable, low‑cost AI‑driven supply‑chain decisions.
AIModel StabilityReinforcement Learning
0 likes · 22 min read

JD Explore Research Institute’s Visual and Multimedia Lab team “Silver‑Bullet‑3D” secured top positions in the 2021 SAPIEN ManiSkill Challenge by excelling in both imitation‑learning and rule‑based tracks, showcasing cutting‑edge computer‑vision and robotic‑arm control technologies that earned them international recognition.
AI competitionComputer VisionRobotics
0 likes · 5 min read

The article surveys three recent approaches—Transition State Clustering, Sequential Windowed Inverse Reinforcement Learning, and Deep Discovery of Continuous Options—that automatically segment long‑horizon surgical‑robot demonstrations into sub‑tasks, learn hierarchical policies from limited data, and achieve markedly higher success rates on da Vinci cutting, tension, and needle‑picking tasks.
Reinforcement LearningRoboticshierarchical learning
0 likes · 18 min read
