Anindulsz

Don't be afraid to be different
Let's learn crypto together, one insight at a time
Contributor @AxisRobotics

[Anindulsz @Tind_220]

@axisrobotics The baseline figure is higher than the previous conference result, and the model used is more advanced

[Anindulsz @Tind_220]

@Chunkyweb3 Follow the plan🫳

[Anindulsz @Tind_220]

A research lab just trained robots with ZERO real world data. pure simulation only

Result? 79.2% success on real robots. More than double the model trained on real world data.

This is exactly the thesis axis robotics has been building on

Simulation isn't plan B, it IS the plan

[Axis Robotics @axisrobotics]

https://t.co/S6oZVikS2v

[Anindulsz @Tind_220]

Bear has no father figure

bera never learn

[Michael Saylor @saylor]

Tranquilize the ₿ears.

[Anindulsz @Tind_220]

@0xCARELESSBOY @axisrobotics @AxisRoboticsID it hugeee

[Anindulsz @Tind_220]

@boybrands_ @axisrobotics @AxisRoboticsID Keep clicking bro

[Anindulsz @Tind_220]

founder axis robotics dropped this in April

45k users, 500k trajectories just 1 month in. "a strong start, but a long road ahead"

fast forwad to today:
> 2.3M trajectories
> $12M seed round closed
> Real commercial deliveries shipped

the road is being built. FAST

[ChrisF(✱,✱) @chris_anm01]

Axis 2026 Roadmap & Outlook

Our Data Collection Platform(https://t.co/1Kr7M9TTMc) has been live for one month — 45K registered users, 500K data trajectories collected. A strong start, but a long road ahead.

Two objectives drive everything this year: close the full product loop from data generation to model training, and scale an open ecosystem globally.

Here's the breakdown 👇

[Anindulsz @Tind_220]

@Chunkyweb3 it's hugeee, wen yah

[Anindulsz @Tind_220]

@xStocksFi site now is back again yihaaa

[Anindulsz @Tind_220]

why @xStocksFi site always like this

oh no, my daily login will break if it continues like this

[Anindulsz @Tind_220]

Axis points already live!

i got 600pts rn, now check you points how much you get ?

for people who want join, you still can join: https://t.co/rLsnA00c3c

[Axis Robotics @axisrobotics]

The Axis Point System is LIVE.

Every valid data contribution is recorded, quantified, and reflected in your Points—so real contributors are recognized for honest, high-quality work that advances robotics.

Check your Points → https://t.co/o543a0DNbW

How it works ↓

[Anindulsz @Tind_220]

The number is going UP only

Robot intelligence is going onchain

[Anindulsz @Tind_220]

Interesting from Axis robotics

Old Task: 100 demos=working policy
New randomized task: 100 demos=failed
500 demos:working policy

Why? new task are harder to memorize, the harder to memorize > the more generalizable the robot.

That's how you build real intelligence

[Axis Robotics @axisrobotics]

Axis Weekly

Last week, we made progress across the full robotics data loop, including task generation, simulation infrastructure, model training, and failure recovery.

Key updates:

- Task generation: We improved TaskGen with better automatic checker generation, stronger multi-embodiment support, and more efficient domain randomization to scale task diversity with less manual design effort.

- Simulation infra: We continued improving MuJoCo verify/replay and scene-variant workflows, including fixes across data collection, multi-asset scenes, repeated loading/downloads, initial states, teleoperation, IK, and gripper control.

- Model training: We confirmed that the new randomized tasks are learnable with sufficient data. In our current experiment, 500 demos successfully produced an executable policy, while 100 demos were not enough.

- Failure recovery: We began building a recover-from-failure pipeline to collect and categorize gripper failure and near-failure states during grasping, which will later support more robust recovery policy learning.

A closer look at this week’s progress🧵

[Anindulsz @Tind_220]

@Arsaasky @axisrobotics @AxisRoboticsID long journey

[Anindulsz @Tind_220]

@Chandra8141 @axisrobotics gAxis

[Anindulsz @Tind_220]

@zero0_f @axisrobotics Congrats W

[Anindulsz @Tind_220]

@axisrobotics @AxisRoboticsID more data isn't the answer

The right data is, and axis built an engine that automatically finds what "right" means

by letting the model expose its own blind spots.

Still learning✌️

[Anindulsz @Tind_220]

Everyone assumes robot AI gets smarter from more data. I did too

Then i read Axis robotics latest research, and it completely changed how i think about what "good data" means

What i learned ?👇

[Anindulsz @Tind_220]

@axisrobotics @AxisRoboticsID and they published the actual number

Round 1 community data -> 57.1% success rate
Identified failure -> targeted corrections collected -> Round 2 -> 66.3% success rate

Successful state: 571 -> 663
Scene coverage: 23 -> 26 variant

The loop works