All Episodes
Embodied AI 101 — 401 episodes · page 4 of 5
LabVLA: Bringing Vision-Language-Action to the Chemistry Lab
Humanoid-GPT: A Foundation Model for Zero-Shot Humanoid Control
CHORUS: Decentralized Multi-Robot Collaboration with a Single Shared VLA Model
RISE: Self-Improving Robot Policy with Compositional World Model
EmbodiedOneVision: Interleaved Vision-Text-Action Pretraining for General Robot Control
Robix: A Unified Model for Robot Interaction, Reasoning and Planning
Robotic World Model: Learning to Simulate for Robust Robot Control
AnchorWorld: Embodied Egocentric World Simulation with View-based Evolution Customization
ArtiFixer: Few-Step Diffusion for 3D Scene Reconstruction
Deployment-Time Memorization in Foundation-Model Agents
Adversarial Machine Learning: Taxonomy, Threat Models, and Mitigation Strategies in Deep Neural Networks
SoCRATES: Evaluating LLM Mediators in Conflict Scenarios
Unembedding Matrix as a Feature Lens: Unlocking Better Text Embeddings
LeanMarathon: Autonomous Formalization of Math Proofs on Erdős Problems
Deep Research Agents: Survey and Roadmap for Autonomous AI Research
Cosmos 3: Omnimodal World Models for Physical AI
Humanoid-GPT: GPT-Style Transformer for Zero-Shot Dynamic Humanoid Control
Bending Paper, Shaping Dexterity: The Robotic Origami Challenge
GraspGen-X: A Foundation Model for Zero-Shot 6-DoF Grasping
When Does Deep RL Beat Calibrated Baselines?
Training Deep Networks as Random Effects: An Optimization–Inference Duality
Generative Depth Supervision for Embodied Vision-Language Models
PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding
LT2: Linear-Time Looped Transformers
One Learning Rate Doesn't Fit All: Layerwise Spectral Scheduling for Transformers
SimToolReal: Procedural Tool Generation and a Universal Objective for Zero-Shot Tool Manipulation
Robometer and the Future of Robotic Reward Modeling
Qwen-VLA: A Generalist Vision–Language–Action Robot Model
EXPO-FT: Sample-Efficient Reinforcement Learning Fine-Tuning for Vision-Language-Action Models
RoboMeter: Learning Dense Rewards from Successes and Failures
MobileGym: A Controllable, Parallel Sandbox for Mobile GUI Agents
ANY2ANY: Efficient Cross-Embodiment Transfer for Humanoid Whole-Body Tracking
TriSplat: Feed-Forward 3D Reconstruction with Triangulated Meshes
MIKASA-Robo-VLA: A Memory-Intensive Benchmark for Vision-Language-Action Robotics
PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation
Bimanual Pegboard Manipulation: A Benchmark for Vision-Language-Action Models
FutureSim: Replaying Real-World Events to Evaluate AI Forecasting Agents
AgentFloor: A Benchmark for Long-Horizon Agent Planning
AlexNet: The Deep Convolutional Network That Transformed Vision
A Few Useful Things to Know About Machine Learning
SimToolReal: A Universal Dexterous Tool-Use Policy
Mimic-Video: Learning Physics Priors from Web-Scale Video for Robot Dexterity
Deep Residual Learning for Image Recognition (ResNet)
Attention Is All You Need – The Transformer Revolution
NVIDIA Cosmos: World Foundation Models for Physical AI
LATENT: Teaching a Humanoid to Play Tennis from Imperfect Data
CollabVR: Collaborative Video Reasoning with Vision-Language and Video Generation Models
World Action Models: The Next Frontier in Embodied AI
Training a Whole-Body Control Foundation Model
DexJoCo: A Unified Benchmark for Task-Oriented Dexterous Manipulation
MMSkills: Building Multimodal Skill Libraries for Visual Agents
PhysBrain 1.0 VLA (TwinBrainVLA): Dual-Brain Vision-Language-Action with Physics-Grounded Learning
MolmoAct2-LIBERO: An Open Vision-Language-Action Model for Robotics
SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Diffusion Transformers
WildClawBench: A Real-World, Long-Horizon Benchmark for AI Agents
MCP-Cosmos: Bring Your Own World Model
OpenAI o1: Teaching LLMs to Think Slow and Deep
The Llama 3 Herd of Models
LATENT: Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data
AnyFlow: Any-Step Video Diffusion for Predictive World Modeling
# Robotics: The Endgame
Claw-Eval: Toward Trustworthy and Transparent Evaluation of Autonomous Agents
LIBERO-Para: Paraphrase Robustness in Robotic Manipulation
YOR: Your Own Mobile Manipulator for Generalizable Robotics
EgoSim: Egocentric World Simulator for Embodied Interaction Generation
Accelerating Video World Models: From Generative Videos to Real-Time Simulators
From Tokens to Thoughts: Continuous Latent Reasoning in Large Models and Robot Control
CaP-X: Coding Agents for Physical eXecution
DoRA: Weight-Decomposed Low-Rank Adaptation
AI Model Collapse: What Happens When AI Trains on Its Own Outputs
PhAIL: Benchmarking Vision-Language-Action Models on Real-World Bin-Picking
Co-training Large Behavior Models: Data Modalities and Training Strategies for Robot Manipulation
HyDRA: Hybrid Memory for Dynamic Video World Models
# WildWorld: Dynamic World Modeling with Actions and Explicit State
Omni-WorldBench: Evaluating Interactive 4D World Models
SIMART: From Static Meshes to Sim-Ready Articulated Models
EgoSim: An Egocentric World Simulator for Embodied Interaction
Digit's New Motor Cortex: Sim-to-Real RL for Whole-Body Control
EgoNav: Diffusion-Based Humanoid Navigation from Human Egocentric Video
CaP-X: A Code-as-Policy Framework for Robot Manipulation
Embodied Intelligence Breakthrough: Generalist AI’s GEN-1 Robots
CaP-X: LMs' First Physical Exam
AI Model Collapse: The Danger of Training on AI-Generated Data
High-Level Automated Reasoning with Qwen2.5-7B
Co-Training Large Behavior Models: Multimodal Data for Robot Manipulation
HyDRA: Hybrid Memory for Dynamic Video World Models
DexWM: Leveraging Human Videos for Dexterous Robot World Models
World Models in Robotics
SIMART: Decomposing Monolithic Meshes into Sim-Ready Articulated Assets
LeWorldModel: A Stable JEPA World Model from Pixels
World Models for Robots: The Next Big Leap?
Harnessing Long-Running AI in Embodied Systems
HoMMI: Learning Whole-Body Mobile Manipulation from Human Demonstrations
TurboQuant: Redefining AI Efficiency with Extreme Compression
DexWM: Learning Dexterous Object Manipulation from Human Videos
FlashAttention-3: Fast & Accurate Attention with Asynchrony & Low-Precision
When AI Trains on Its Own Output: The Model Collapse Problem
MolmoBot: A Vision-Language Model for Zero-Shot Robot Manipulation
LeWorldModel: Stable End-to-End JEPA from Pixels