
atharva ☆
@k7agar • 12,061 subscribers
your friendly neighbourhood engineer
Shorts
Videos

I thought the tasks they were doing were too simple, so I asked Astra to “pick up the green block and make it upright.” the way it did it left both sujal and me speechless. I didn’t want to believe it, but now I do scaling is all we need. makes me so much more bullish on VLA's in general
atharva ☆36,857 Aufrufe • vor 3 Tagen

MolmoAct2 kinda just works out of the box in an unseen environment and with unseen objects. I find it really crazy that I can just download a model from the internet and plug it into my robot and it can do some tasks at a limited capacity right away. this video is at 2x speed
atharva ☆32,346 Aufrufe • vor 4 Monaten

I tried implementing and benchmarking the π₀ VLA on modal GPU's - I measured latency overhead (HTTP/WS/TCP/QUIC) - measured closed-loop success rates in aloha simulator imo, this is definitely the future of deploying policies on robots more optimizations can get it down to be completive in realtime. network is the bottleneck but can be solved, this unlocks so many applications. I am hoping to try doing this on hardware next. github in commets
atharva ☆36,387 Aufrufe • vor 5 Monaten
Keine weiteren Inhalte verfügbar