Hacker News - Newest: "robot"

This is a write-up of a system I designed at a previous job. It's a natural language interface for industrial robots. The demo runs MuJoCo in the browser via WASM.

The demo uses a free-tier LLM API with rate limiting, so it might be slow or unavailable under load, but feel free to check out the code and hook up your own inference service: https://github.com/Encrux/llm_trajectory

Comments URL: https://news.ycombinator.com/item?id=47674502

Points: 3

# Comments: 0

OpenAI's vision for the AI economy: public wealth funds, robot taxes, and more

evo_9 — Mon, 06 Apr 2026 23:55:34 +0000

Article URL: https://techcrunch.com/2026/04/06/openais-vision-for-the-ai-economy-public-wealth-funds-robot-taxes-and-a-four-day-work-week/

Comments URL: https://news.ycombinator.com/item?id=47668974

Points: 4

# Comments: 0

I Woke Up Not Knowing Who I Was – Building an AI Robot Brain [video]

cobbzilla — Mon, 06 Apr 2026 13:29:22 +0000

Article URL: https://www.youtube.com/watch?v=7T3ogtB5YS0

Comments URL: https://news.ycombinator.com/item?id=47660648

Points: 1

# Comments: 0

Reviving a Robot

kukanani — Mon, 06 Apr 2026 13:11:53 +0000

Article URL: https://allevato.me/2026/04/03/reviving-a-robot

Comments URL: https://news.ycombinator.com/item?id=47660464

Points: 1

# Comments: 0

Show HN: We built a camera only robot vacuum for less than $300 (well almost)

indraneelpatil — Mon, 06 Apr 2026 05:08:20 +0000

Article URL: https://indraneelpatil.github.io/blog/2026/robot-vacuum/

Comments URL: https://news.ycombinator.com/item?id=47657204

Points: 108

# Comments: 56

In Japan, the robot isn't coming for your job; it's filling the one nobody wants

rbanffy — Sun, 05 Apr 2026 22:31:16 +0000

Article URL: https://techcrunch.com/2026/04/05/japan-is-proving-experimental-physical-ai-is-ready-for-the-real-world/

Comments URL: https://news.ycombinator.com/item?id=47654620

Points: 199

# Comments: 241

Robot Umpires Have Arrived–and They're Making Baseball Players Shrink

coloneltcb — Wed, 01 Apr 2026 19:12:19 +0000

Article URL: https://www.wsj.com/sports/baseball/mlb-robot-umpires-abs-shrinking-65eaacf5

Comments URL: https://news.ycombinator.com/item?id=47605177

Points: 1

# Comments: 0

Show HN: An open source rubber ducking robot, also a joke

jftesser — Wed, 01 Apr 2026 16:48:01 +0000

Article URL: https://duck-duck-duck.edges.ideo.com/

Comments URL: https://news.ycombinator.com/item?id=47603339

Points: 3

# Comments: 0

Atombite.ai Deep Dive: Building a Takeout Packing Robot Is Harder Than You Think

emmanol — Wed, 01 Apr 2026 08:22:35 +0000

In 1997, IBM's Deep Blue beat Garry Kasparov at chess. In 2016, DeepMind's AlphaGo defeated Lee Sedol at Go. Yet, in 2026, if you ask a $100,000 robotic arm to pick up a crushed paper bag, place a hot soup container inside without spilling it, and staple a receipt to the top, it will likely fail catastrophically.

This is Moravec's Paradox in action: the high-level reasoning tasks that humans find difficult are trivial for computers, while the low-level sensorimotor skills we take for granted are computationally intractable .

AtomBite.AI is an artificial intelligence application company building the "AtomBite Brain"—a foundation model for flexible manipulation in commercial robotics. While the industry has poured over $7.2 billion into humanoid robots focused on locomotion and rigid tasks , AtomBite.AI is focused entirely on the cognitive bottleneck of the grasping problem in chaotic environments like commercial kitchens.

The Physics of Infinite Degrees of Freedom

"Manipulation is the hard problem we need to solve to make humanoid robots useful, not locomotion." — Bob McGrew, former OpenAI Chief Research Officer

The fundamental difference between industrial automation and embodied AI lies in the physics of the objects being manipulated. A chess piece or a car chassis is a rigid body. It has exactly six degrees of freedom (three for translation, three for rotation). If a robot's vision system identifies the coordinates of a rigid object, the inverse kinematics required to grasp it are a solved mathematical problem. Industrial robots like the FANUC M-10iA can achieve a repeatability of ±0.03 millimeters .

However, a paper takeout bag is a deformable object. It possesses theoretically infinite degrees of freedom. When a robotic gripper makes contact with the bag, the bag's geometry changes instantly. The state estimation required to track these deformations in real-time exceeds the capabilities of traditional physics engines.

Furthermore, the takeout packing process involves extreme material diversity. In a single 30-second window, a robot must handle a rigid plastic lid, a deformable paper bag, a liquid-filled cup with shifting center of mass, and a flimsy paper receipt. Traditional robotic systems require reprogramming if an object's size, shape, or even color changes slightly . In a commercial kitchen, no two orders are identical.

The Dual-Model Architecture Solution

To solve the latency and compute constraints of flexible manipulation, AtomBite.AI developed a Dual-Model Architecture that separates high-level reasoning from low-level execution.

Running a massive vision-language-action (VLA) foundation model at 50Hz to control motor torques is economically and computationally unviable for a commercial kitchen. Instead, the AtomBite Brain splits the cognitive load.

The Foundation Model (System 2) acts as the slow, deliberate reasoning engine. It processes the chaotic visual scene, identifies the edge cases (e.g., "the soup container lid is slightly ajar"), and generates a high-level semantic plan.

The Edge AI (System 1) acts as the fast, reactive motor cortex. It translates the semantic plan into high-frequency, low-latency motor control commands. If the edge model encounters a state it cannot resolve—such as the paper bag tearing during the lift—it instantly queries the foundation model for a new strategy.

"We realized early on that building a better mechanical hand wasn't going to solve the problem," explains Dr. Dong Wang, CEO of AtomBite.AI and former CTO of Meituan Delivery. "The bottleneck is entirely cognitive. You need a brain that can reason about the physical properties of a wet paper bag in milliseconds, without relying on pre-programmed waypoints."

Hacker News - Newest: "robot"

Waymo's Robot Car Testing Ends in NYC After Permits Expire

OpenAI's vision for AI economy: public wealth funds, robot taxes, 4-day workweek

FusionCore ROS 2 sensor fusion replacing deprecated robot_localization

Where Is the Midjourney for Robot Simulations?

A soft robot has no problem moving with no motor and no gears

China trailing behind in robot density

I Hired a Robot and Made Him Journal

Waymo's Robot Car Testing Ends in NYC After Permits Expire

The Open-Source Recipe for Teaching a Robot to Fold Your Clothes

An AI robot in my home

Show HN: Letting an LLM write robot programs

OpenAI's vision for the AI economy: public wealth funds, robot taxes, and more

I Woke Up Not Knowing Who I Was – Building an AI Robot Brain [video]

Reviving a Robot

Show HN: We built a camera only robot vacuum for less than $300 (well almost)

In Japan, the robot isn't coming for your job; it's filling the one nobody wants

Robot Umpires Have Arrived–and They're Making Baseball Players Shrink

Show HN: An open source rubber ducking robot, also a joke

Atombite.ai Deep Dive: Building a Takeout Packing Robot Is Harder Than You Think

The End of the "I Am Not a Robot" Box