Peter Okiokpa

Robotics has three hard problems: semantics, planning, and real-time control. Language models cracked the first one. Planning got better fast. Control in dynamic environments is still a work in progress.

If you’re building a robot right now, this is for you. @tnkrdotai Build Hours. Every Tuesday. Same time. Bring your questions, your half-finished builds, your “does anyone know why this isn’t working” moments. Real feedback from builders who care. Link in comments.

Asimov's BOM now has retail prices next to most parts. manual.asimov.inc/v1/bom

@KuphDev @lukas_m_ziegler Oh, I do have my eye on the openarm 101 it uses CAN bus-based motors (more precise, lower latency) compared to the SO-101's serial servos. You can get more deets about it here roboticscenter.ai/learn/robot-ar…

@peterokiokpa @lukas_m_ziegler Are there any other robot arms you have your eye on?

100% open-source robotic arm! 🦾 Seeed Studio released reBot-DevArm, a robotic arm project lowering the barrier to learning robotics... or how people call it these days: physical AI. Everything is open-sourced. Hardware blueprints include sheet metal and 3D printed parts. Detailed BOM covers every screw with purchase links. Software and algorithms include Python SDK, ROS1/2, Isaac Sim, and LeRobot. Let's have a look at robot specs. So it has 1.5 kg payload, 650 mm max reach, 4.5 kg weight, and less than 0.2 mm repeatability with 6 DoF plus gripper. This is true open source for robotics. When every screw, CAD file, motor driver, and algorithm is freely available, desktop robotic arms become accessible to students, researchers, and developers worldwide. ‼️ Start your career in robotics today: github.com/Seeed-Projects… ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com

Dear young Nigerian, Take a closer look, you may not be where you wish to be yet, but you are definitely not where you used to be. That’s PROGRESS. Keep your hope and dreams alive. You'll make. Not in your own strength, but with God by your side. Your best days are ahead of you!

@KuphDev @lukas_m_ziegler Same here

@lukas_m_ziegler @peterokiokpa I’m pretty hyped about this! Pretty sure it’s gonna be my next robot arm. The SO-101 has been fun, but feels more like a “toy” than something I can do advanced projects with.

Raspberry Pi Zero 2 Wを使ったGPSの実験。外に持ち出すのが楽しみ。電源は小型のモバイルバッテリーを使います。

if you're new to actuators, study these companies: strain-wave reducers (rotary transmission): - Harmonic Drive Systems, Japan - Leaderdrive, China - Green Harmonic, China planetary roller screws (linear transmission): - Rollvis, Switzerland - Schaeffler / Ewellix, Germany - SKF, Sweden frameless BLDC motors: - Kollmorgen, US - maxon, Switzerland - TQ-RoboDrive, Germany - CubeMars, China bearings: - THK, Japan - IKO, Japan torque sensors: - ATI Industrial, US - FUTEK, US complete actuator modules: - Hyundai Mobis, Korea - Apptronik, US - Unitree, China - AgiBot, China encoders: - RLS / Renishaw, UK - Celera Motion, US Japan, China, Switzerland, Germany dominate let me know who im missing

🤖🍇 Robotics & World Model Reading Club 06 Recap—SF0502 @saturdayrobotic Video World Models as Simulator & Policy—Keynote from @tongzhou_mu (@RhodaAI), cohost @junfanzhu98, @aurorafeng_01 Robotics is no longer about learning actions—it’s about selecting actions from predicted futures. 🎬 Two Roles of Video World Models - Simulator → learn physics from data, generate experience - Policy → drive decisions via video-conditioned action generation Goal: inject physical common sense from web-scale video into control. 🧪 I. Video Models as Learned Simulators 1) Data Synthesis (DreamGen, GR00T) Pipeline = RAG + rollout + IDM labeling - Prompt → retrieve similar robot videos → generate new task rollouts - Inverse Dynamics Model (IDM) → pseudo actions → train policy ✅ Cheap, scalable, safe edge cases ❌ Open-loop → hallucination + error accumulation ⚠️ Insight: IDM is NOT the bottleneck → inverse mapping is easy; forward world modeling is hard → works with teleop / eval / random data; generalizes across robots 2) Inference-Time Planning (V-JEPA2) - Action-conditioned video model -Sample action sequences → rollout in latent → score vs goal - Replan iteratively (receding horizon) ✅ Test-time scaling (more samples = better plans) ❌ Heavy compute vs real-time control 👉 Pipeline: Policy eval → prune → planning 3) Policy Evaluation (Veo, Ctrl-World) - Use video model as simulator for scoring trajectories - Acts as action filter / value proxy ✅ Unlimited rollouts (vs traditional sim limits) ❌ Less accurate than physics engines 🚨 Not real-time → offline selection before planning 🤖 II. Video Models as Policy 4 Paradigms (1) Joint Video + Action Generation (DreamZero, GR1/2) - Diffusion / flow matching over video+action - Shared denoising → cross-modal reasoning ⚠️ Open: pretrained video ≠ pretrained for action (2) Representation → Action (VPP, Video Policy) - Video model = feature extractor - Small diffusion policy = action head ⚡ Fast inference ⚠️ Partial denoising = control authority allocation - none → action head decides - full → video dominates - partial → shared decision boundary (3) Open-loop Generation (UniPi) Generate full future video → IDM → actions ✅ Uses video model as-is ❌ Plan fixed → no reaction → brittle (4) Closed-loop Generation (DVA, LingBot) Generate → act → replace with real observation → repeat ✅ Continuous grounding → avoids hallucination ❌ Requires causal modeling + heavy infra 🔥 Core insight: Video model ≠ decision maker → it proposes futures → policy = selecting among futures via translation/scoring 🧠 System-Level Truths - Failure ≠ video problem → usually translation / constraint / IDM issue - Action space is task-dependent (position vs others) - Closed-loop = continuous alignment with reality ⚙️ Deployment Reality Infra > model tweaks (latency, kernel fusion, precision) 100 robots ↔ 1000 GPUs? edge vs cloud tradeoff Data unclear: UMI / egocentric promising but not converged Perception bottleneck: camera latency, resolution, depth ⚠️ Fundamental Tensions Latent vs Pixel → latent efficient but may drop task-critical info → representation choice caps capability RL Warning → learned simulator ≠ ground truth → RL will exploit model bias (simulator hacking) Tactile vs Vision → easy to add, but no internet-scale data → video dominates 🚀 Emerging Directions - Diffusion distillation → faster generation - Auto-regressive diffusion transformers - Video models as simulator + policy + value function + data engine 👉 A unified computational primitive 🧩 Video world models are not just better perception—they redefine control: From “predict → act” → to “generate futures → select actions.” The bottleneck has shifted: ❌ Not model capability ✅ Grounding decisions into real-world constraints

You can build, train, and customize a humanoid robot.

The most underrated thing about what @tnkrdotai is building every robot that gets built and connected becomes a node in a fleet that fleet collects real world data that data trains better models that's the flywheel. builders → robots → data → intelligence worth paying attention to

A daily window into the universe. 🌌📟 Daily Cosmic Explorer by Young brings NASA’s Astronomy Picture of the Day, image credit, date, and a short cosmic explanation to your ePaper display. Built with our SenseCraft HMI, no code. Best for reTerminal E1004. Explore the design 👉 sensecraft.seeed.cc/hmi/template/1…

There should be an electronics conference that is also a rave party!

Making it easier for people to build robots

The biggest challenge in AI today isn't building a model — it's making that model work on a device with low latency and high efficiency for real-world projects. We’ve partnered with @Qualcomm Academy to bridge that gap. Using @EdgeImpulse and Arduino UNO Q, this certification gives you the blueprint to turn raw data into a fully functional edge AI application. Ready to get certified? Enroll now to master the UNO Q's "dual-brain" architecture and develop systems that think as fast as they act: academy.qualcomm.com/course-catalog…

ラズパイと電子部品を配線してる人の座り方

The Physical AI Robotics GR00T‑X Embodiment Sim dataset has surpassed 10 million downloads on @HuggingFace. 🥳 A huge shoutout to the global research and developer community exploring the future of embodied AI and robotics with this open dataset — you made this milestone possible. 📥 Try it on Hugging Face 👉 nvda.ws/3Qv64Ul

Really cool