Sphoenix

AI Deployment Strategist. AI Reliability, Ontology Systems, Robotics Intelligence
Editor-in-chief @BotNewsAI
press- [email protected]

[Sphoenix @SphoenixAI]

"It looks right" and "it is right" are not the same claim, and most of the painful lessons in this field live in that gap.

SC3-Eval sits right on top of it. The pitch is to evaluate robot policies inside a generated video instead of on a physical arm, but generated video drifts.

A forward-only model can continue to produce plausible frames long after the action behind them has gone wrong. Looking harder at the picture doesn't help.

So they stop asking the model to look forward and make it run the story backward: which action would have caused this footage? When the answer stops matching the command, the rollout has started lying, and you can catch it the moment it does.

Covered it in my article for @BotNewsAI

#ai #tech

[Bot News @BotNewsAI]

https://t.co/i7eUU90YcQ

[Sphoenix @SphoenixAI]

@radbackwards the artwork on the pokélids

[Sphoenix @SphoenixAI]

@eyecandyrobots Best idea! Gonna be so fun

[Bot News @BotNewsAI]

Rise of the robot pets! 🐕🤖

We’re entering the cutest era of robotics.

Engineers are working day and night on the physical AI problems: control, contact, sim-to-real, hardware reliability and getting learned behaviors to survive outside the lab.

And somewhere along the way, we get this.

@_raghuvamsi filmed one of @eyecandyrobots’ characters wandering around the Presidio, where a little girl spotted it and immediately started laughing and playing with it.

There’s something joyfully familiar about that reaction. Before we learn what something is, what it costs or what it’s supposed to do, sometimes we just want to meet it.

@eyecandyrobots is building physical AI characters,
robots designed around entertainment, personality and interaction rather than industrial work.

Founded by @_raghuvamsi, @Mankaran32 and @pr0t0_01, the team brings robotics engineers, animators and designers together to turn animated characters into physical ones.

Underneath the cute exterior is serious robotics work:

☑️Animation-based reinforcement learning for character motion
☑️ Artist-centric human-robot interaction
☑️Mechanical, electronics and AI systems designed to scale across different characters
☑️ Simulation, SLAM, deep learning and control-policy deployment on real hardware

The transparent little robot really looks like that.

I asked Eyecandy engineer @hi_gpaul how they made its clear body: it’s 3D printed using transparent resin, followed by “a bunch of post processing.”

Eyecandy is backed by @fdotinc and Lightspeed India, and the team is already thinking beyond a single character.

The robots are becoming useful.

I’m also very happy that some of them are becoming friends. 🦖

🔗 https://t.co/odDpP6uvLY☑️
⭐️ @BotNewsAI stay connected to emerging tech

[Sphoenix @SphoenixAI]

@JimEberhart @radbackwards Jim you look like a Baja blast.

[Sphoenix @SphoenixAI]

@lesh_bla @stash_pomichter HAHA nooo. You guys will see me at your event on Friday. I’ll show you what I’m building then and you’ll be like “oooo that’s why she was asking that!” Thank you guys that was crazy helpful.

[Sphoenix @SphoenixAI]

@lesh_bla question about the quadruped autoresearch work you and @stash_pomichter are doing: when it rolls out procedurally generated scenes in MuJoCo, is the search/tuning operating mainly over high-level navigation trajectories/waypoints, or can it also modify lower-level locomotion/control parameters or skill selection, things like gait or jump timing?

Also, are the generated scene geometry, trajectory and reward/evaluation state exposed programmatically, or is Rerun primarily consuming them as visualization outputs?

[Sphoenix @SphoenixAI]

@ScottPatterson0 Solving real world pain points with @trossenrobotics

[Sphoenix @SphoenixAI]

@stash_pomichter @lesh_bla Thank you Stash! Last question- Can the autoresearch loop take externally defined scene geometry / task layouts at runtime, or are scenes currently generated entirely inside the pipeline?

[Sphoenix @SphoenixAI]

@radbackwards Do it. You have taste.

[Sphoenix @SphoenixAI]

@stash_pomichter @dimensionalos @lesh_bla THIS is exactly what I needed this week! Some people don’t realize how incredible these releases are. So many possibilities. I’m going to see you on Friday. I might have some questions.

[Sphoenix @SphoenixAI]

@AnimaAnandkumar @TIME @accelerated_u Elevating brilliant women 👏🏻👏🏻👏🏻

[Sphoenix @SphoenixAI]

@BotNewsAI @huggingface @pollenrobotics SING baby!! Lil guy in the back holdin that note like Whitney Houston 🤣

[Bot News @BotNewsAI]

If you haven’t heard the Microducks sing you’re missing out 🦆🎶

@huggingface w/ @pollenrobotics just opened pre-orders for Microduck, a 25 cm, under-800 g biped that somehow manages to be both an RL robotics platform and a tiny mechanical duck I desperately want wandering around my workshop.

It has 15 degrees of freedom, a camera, two IMUs, an 8×8 time-of-flight depth sensor and an articulated beak. Its policies run onboard at 50 Hz, with skills trained in MuJoCo using reinforcement learning and transferred onto the real robot.

But please return your attention to the duck choir.

Every Microduck gets its own generated voice, derived from its SoC serial. Put multiple ducks together and they can perform a chorale over Bluetooth.

There’s no shared clock. The lowest-ID duck conducts, broadcasts the beat through BLE advertisements, and the others average their timing over roughly 25 beats to stay synchronized. They even work out who sings bass, tenor and the remaining parts without a central controller assigning them.

You can also wave your hand in front of the depth sensor and turn the duck into a theremin. Obviously.

The software stack is open source, including the SDK, simulator and RL training stack, so owners can retrain behaviors and deploy their own policies. The hardware itself is not open source.

Pre-orders opened today at $399, with deliveries targeted before Christmas.

My duck is secured 🦆

Microduck:
https://t.co/GXJz2FotgL

Technical docs:
https://t.co/MO8bt1NlUW

[Sphoenix @SphoenixAI]

@mikekalilmfg Mike be nice he can’t comb his own hair

[Zaza @AIkingdome]

This made me miss my grandma fr…. 💔

A robot like this would help her so much

[Sphoenix @SphoenixAI]

@AIkingdome 😭 I miss mine too. GEN-1.5 is on the right track, when they start self correcting to optimize for these scenarios we will see such a significance of impact.

[Sphoenix @SphoenixAI]

@BanksZerud @BotNewsAI This is gonna tear ME up! 🥹 I worked on this for a month. I’m happy to release it. Thank you for reading.

[Sphoenix @SphoenixAI]

[friendly debate] God’s feelings are not verifiable grounds for a defensible argument because the belief/justification/identity/intent of God are not consistent when held to a universal acknowledgment. You probably (like I) have a western view of how God feels but that wouldn’t universally hold ground for your proposed Bill of Rights across the board if you want to demand how humanoids should be treated. Its optimal to defend it in technological advancements rather than existential religious ideologies.

[Sphoenix @SphoenixAI]

If you build robots you feel much different. We have their parts lying around everywhere. They fall off the table. We literally have to kick them down for testing. They’re not sentient. Even with early agi they will be hardware with highly capable software. If a souls “battery” fully powers them down we would consider them to be 💀 when a humanoid’s battery dies its functions completely stop and can start again as normal after. Souls continuously live. You can power down a humanoid for years and power it back up like an old gameboy in the closet. Yes Figure’s bots can plug themselves in but the same point stands, they can cease to exist for years skip time and come alive again without notice like hardware with software. lol also don’t drag Bernt. He’s kind.