Embodied AI 101

Shaoqing Tan

Embodied AI 101 tracks the latest research at the intersection of AI and physical intelligence.

Robotics, manipulation, world models, and the path from digital intelligence to embodied agents — each episode breaks down recent papers and advances in the field.

We examine the methodology, real-world implications, and path to deployment. Covering robot learning, perception, control, and the fundamental challenge of grounding intelligence in the physical world.

New episodes daily.

Play

Open in

107 episodes
Avg 26 min
English

Episodes 107

Newest Oldest Longest

All20 Unplayed20 In progress0 Played0

Counted on this page — what you have heard stays on this device, so it is not something the list can be paged by.

Today · 12 min
Show-Harness: Can Small VLMs Replace Frontier Models in Robot Control?
Show-Harness fine-tunes a lightweight VLM (e.g., Qwen3.5-2B) in hours to serve as a capable local robot agent, eliminating the need for large models at inference time. This approach makes VLM-based robot intelligence practical for on-device deployment wit
- Transcript
Yesterday · 13 min
MolmoBot: Can Simulation Alone Deliver Zero-Shot Manipulation?
MolmoBot leverages large-scale simulation to enable zero-shot manipulation transfer to real robots without task-specific training data. This advances the sim-to-real generalization frontier for vision-language-action models in manipulation.
- Transcript
Yesterday · 13 min
VLA-Δ: Can World-Change Supervision Improve Robot Manipulation?
VLA-Δ trains robot policies on how the physical world changes rather than raw actions, matching state-of-the-art on LIBERO and RoboTwin benchmarks and outperforming π0.5 on real robots. This world-change modeling paradigm offers a novel perspective on gro
- Transcript
Wednesday · 12 min
Axis Robotics Browser Simulation: Does Sim-to-Real Manipulation Work?
A browser-based teleoperation interface enables collection of robot arm demonstrations in simulation that transfer to real hardware (e.g., Franka) via sim-to-real, achieving high success rates with mixed sim and few real demos. This lowers the barrier to
- Transcript
Monday · 12 min
HexaShell: Can Debugging Code Generalize RGB Robot Manipulation?
HexaShell treats robot control as a coding task where an agent writes reusable tools from RGB camera observations, achieving 39.7% success across 42 tasks on the RoboDojo benchmark — a new SOTA. This matters because it demonstrates that code-as-action abs
- Transcript
Monday · 12 min
Niantic Spatial Gaussian Splats: Better Sim-to-Real Robot Training?
Niantic Spatial released high-fidelity 3D Gaussian splat scenes with aligned meshes and collider-ready geometry for warehouse robot simulation and policy training. By combining photorealistic appearance with physically accurate collision geometry, this ap
- Transcript
Monday · 12 min
Gemini 4 Argon for Robotics: What the Benchmarks Actually Show
Google released Gemini 4 Argon, a physical reasoning model that scores 87% on the ERQA embodied-reasoning benchmark using pure reasoning, outperforming other frontier models. Argon controls robots not through raw float trajectories but via a code-as-polic
- Transcript
Sunday · 12 min
GreenWAM: Can One World-Action Model Learn Across Robot Bodies?
GreenWAM is an open-weights world-action foundation model that jointly predicts robot actions and selects them across 20 robot families within a single unified model. Its cross-embodiment generalization and open release make it a significant step toward s
- Transcript
Sunday · 14 min
3DCodeVerse: Can Code Generation Build Better Worlds for Robotics?
3DCodeVerse introduces a model trained on 2M programs that generates static and articulated 3D objects, scenes, and entire worlds represented as executable code. The code-as-representation approach for 3D world generation is a novel contribution to spatia
- Transcript
Saturday · 12 min
Motion- and Gravity-Aware 3D Reconstruction: Do Offline Priors Help?
A new feed-forward 3D reconstruction method injects optical flow and gravity priors directly into the network to robustly handle dynamic scenes and motion blur without any additional inference cost. This advances camera-based spatial intelligence for embo
- Transcript
Saturday · 13 min
ASPIRE: Can Code Refinement Make Robot Manipulation Generalize?
NVIDIA's ASPIRE framework combines GPT-6 Astra with continual skill refinement to achieve saturated performance (99–100%) on LIBERO, Robosuite, and BEHAVIOR-1K simulation benchmarks, with open-sourced code targeting real-world manipulation transfer. The r
- Transcript
October 16 · 13 min
UAV-ESPrune: Can Visual Token Pruning Preserve UAV Reasoning?
Low-altitude unmanned aerial vehicles (UAVs) require compact multimodal inference for perception, environment understanding, and planning-oriented reasoning. Vision-language models (VLMs) provide a shared interface for these tasks, but long visual-token s
- Transcript
October 15 · 30 min
Gemini Robotics 2: How General Is Its Whole-Body Robot Control?
Google DeepMind's Gemini Robotics 2 demonstrates whole-body control, fine-grained dexterity, and embodied reasoning across multi-step real-world manipulation tasks. It highlights the growing importance of rich spatial and environmental data for scaling ro
- Transcript
October 15 · 13 min
RoboJEPA to Long-WAM: Do New Embodied AI Models Generalize?
A cluster of new releases including RoboJEPA (latent world models spanning 12 robot embodiments), UNITAS (3D-native world-action model), and DreamTrue (failure prediction via robot world models) collectively push forward the frontier of embodied world mod
- Transcript
October 15 · 29 min
VLA-ACL: Can Token Pruning Speed Up Robot Manipulation?
VLA-ACL (Action Consistency Learning) learns a lightweight visual token pruning policy through action-level supervision while keeping the base VLA model entirely frozen. The training objective encourages actions produced from pruned visual contexts to rem
- Transcript
October 15 · 28 min
Juno: Do Predictive Latents Make VLA Manipulation More Robust?
Juno introduces a unified framework built around one action-conditioned JEPA that serves as a control-aligned representation backbone, a predictive teacher, and an adaptable dynamics model for VLA models. It addresses three failures: mismatch with embodim
- Transcript
October 15 · 27 min
Can Rephrasing Fix VLA Language Sensitivity Without Retraining?
Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-language models they are built on. A one-word edit can move success by tens of points: pi0.5 turns on a LIBERO s
- Transcript
October 14 · 28 min
Can CriticGPT Replace Hand-Designed Rewards for Robot Manipulation?
Learning scalable reward signals for robotic manipulation remains challenging. CriticGPT uses a multimodal LLM as a critic to provide reward signals for robot manipulation, avoiding the need for manually designed dense rewards or costly human preference a
- Transcript
October 14 · 28 min
Urbex: Can Agentic VLM Search Improve City-Scale 3D Grounding?
Urbex reframes city-scale 3D grounding as an active spatial search problem, using an agentic vision-language model that iteratively queries landmarks, zooms into regions, and renders oblique views to localize targets. The approach outperforms prior baseli
- Transcript
October 12 · 28 min
VPP2: Does Better Video Prediction Improve Robot Manipulation?
World action models (WAMs) have emerged as an important class of generalist robot policies, aiming to transfer video prediction priors to action learning. Finds that existing WAMs frequently fail and proposes improvements.
- Transcript

Showing 1–20 of 107 episodes