we could just remote infer the openpilot model with a local 4090.
and we 3d printed gearboxes
hardware teaches you cool stuff
and we 3d printed gearboxes
hardware teaches you cool stuff
meet McQueen.
an autonomous RC car built to test how pretrained driving models from @comma_ai perform in the real world, using a single RGB camera and remote inference.
0
0
13
350
tactile always seems fun. the same sense we humans have but modeled for machines. would love seeing yall work it out.
I always wondered what would have happened if the robots also had the sense of touch.
So with that curiosity in mind, me and @mahi_zade0712 worked upon the "mission mimosa".
result? policies converge faster and are sample efficient due to the flexi-tac on SO101.
0
0
14
605
representations have always mattered the most but egocentric research highlights it now
when i started with robot learning in general, the most creative cool concept was the study of representation space.
latent world modelling predicts in embedding space which means it basically learns to daydream well
can we call it that
0
0
7
482
what we did
-a dataset of 50 episodes but they dont start from the home pose
-the policy has jerky actions in the initial trajectory of getting up and going towards the tape
-the main action similar to sweeping and making the tape vertical is well learnt
@k7agar @tirth_gada
-a dataset of 50 episodes but they dont start from the home pose
-the policy has jerky actions in the initial trajectory of getting up and going towards the tape
-the main action similar to sweeping and making the tape vertical is well learnt
@k7agar @tirth_gada
0
0
10
356
got our @i2rt_robotics YAMs up and running with gello inspired leader arms for 1/10th of the cost.
@k7agar @tirth_gada
@k7agar @tirth_gada
6
4
38
5,194
theres so much of cheap data collection in India
and so much of money spent to make it useful
and so much of money spent to make it useful
India is quietly becoming a training floor for humanoid robots, with workers filming thousands of first-person hand tasks so AI systems can learn grasping, folding, sorting, and tool use.
This story is really about how the humanoid robot boom still depends on cheap, repetitive chines basic physical skill.
The problem is that robots do not fail on big plans first; they fail on tiny physical details like grip angle, finger timing, slip correction, and object contact.
That kind of knowledge is hard to code and expensive to collect.
These labs capture that missing layer by putting cameras or sensors on people and recording ordinary actions as machine-readable motion examples.
The useful part is not the towel or box itself but the sequence: where the hand starts, how force changes, when fingers adjust, and how the body recovers from small mistakes.
That gives robotics teams supervised data for models that map visual input to physical actions, which is much easier than hand-coding every movement rule.
This is a story about how physical intelligence gets extracted before it gets automated.
---
quasa. io/media/the-hidden-hand-farms-of-india-fueling-the-ai-robot-revolution-with-human-motion
1
0
9
420