๐ค ๐๐๐ฒ๐ฟ๐ ๐ฟ๐ผ๐ฏ๐ผ๐ ๐๐ฎ๐๐ธ ๐ณ๐ฎ๐ถ๐น๐๐ฟ๐ฒ ๐ฐ๐ฎ๐ป ๐ฏ๐ฒ๐ฐ๐ผ๐บ๐ฒ ๐ฎ ๐น๐ฒ๐๐๐ผ๐ป ๐ณ๐ผ๐ฟ ๐๐ต๐ฒ ๐ป๐ฒ๐
๐ ๐ถ๐๐ฒ๐ฟ๐ฎ๐๐ถ๐ผ๐ป.
๐ฆ๐ฒ๐ป๐๐ฒ๐ก๐ผ๐๐ฎ-๐ฅ๐ผ๐ฏ๐ผ๐ฅ๐ฆ๐ uses ๐ฅ๐ฆ๐ to evolve ๐ฎ๐ด๐ฒ๐ป๐ ๐ต๐ฎ๐ฟ๐ป๐ฒ๐๐๐ฒ๐ through physical task feedback, without updating model : 56.83 avg. score, 50.83% success rate, +๐ฎ๐ณ.๐ด๐ฒ ๐ฝ๐ผ๐ถ๐ป๐๐ ๐๐. ๐๐ต๐ฒ ๐๐ฃ๐ง-๐ฒ ๐๐๐๐ฟ๐ฎ ๐ฏ๐ฎ๐๐ฒ๐น๐ถ๐ป๐ฒ (๐ฎ๐ด.๐ต๐ณ)
โถ ๐ฅ๐ฆ๐ ๐น๐ผ๐ผ๐ฝ: Execute โ Diagnose โ Improve โ Validate โ Inherit
โถ ๐๐ ๐ฝ๐น๐ผ๐ฟ๐ฒ๐ฑ ๐ฎ๐ป๐ฑ ๐๐ฎ๐น๐ถ๐ฑ๐ฎ๐๐ฒ๐ฑ: Multi-point EEF prediction and Planning + Feedback Subagents
This achievement reflects the collective efforts of our joint team.
RoboRSI will be ๐ผ๐ฝ๐ฒ๐ป-๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ฑ ๐๐ผ๐ผ๐ป at github.com/OpenSenseNova/โฆ. A comprehensive ๐๐ฒ๐ฐ๐ต๐ป๐ถ๐ฐ๐ฎ๐น ๐ฟ๐ฒ๐ฝ๐ผ๐ฟ๐ detailing our methods and findings is also on the way. Stay tuned!
๐ฆ๐ฒ๐ป๐๐ฒ๐ก๐ผ๐๐ฎ-๐ฅ๐ผ๐ฏ๐ผ๐ฅ๐ฆ๐ uses ๐ฅ๐ฆ๐ to evolve ๐ฎ๐ด๐ฒ๐ป๐ ๐ต๐ฎ๐ฟ๐ป๐ฒ๐๐๐ฒ๐ through physical task feedback, without updating model : 56.83 avg. score, 50.83% success rate, +๐ฎ๐ณ.๐ด๐ฒ ๐ฝ๐ผ๐ถ๐ป๐๐ ๐๐. ๐๐ต๐ฒ ๐๐ฃ๐ง-๐ฒ ๐๐๐๐ฟ๐ฎ ๐ฏ๐ฎ๐๐ฒ๐น๐ถ๐ป๐ฒ (๐ฎ๐ด.๐ต๐ณ)
โถ ๐ฅ๐ฆ๐ ๐น๐ผ๐ผ๐ฝ: Execute โ Diagnose โ Improve โ Validate โ Inherit
โถ ๐๐ ๐ฝ๐น๐ผ๐ฟ๐ฒ๐ฑ ๐ฎ๐ป๐ฑ ๐๐ฎ๐น๐ถ๐ฑ๐ฎ๐๐ฒ๐ฑ: Multi-point EEF prediction and Planning + Feedback Subagents
This achievement reflects the collective efforts of our joint team.
RoboRSI will be ๐ผ๐ฝ๐ฒ๐ป-๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ฑ ๐๐ผ๐ผ๐ป at github.com/OpenSenseNova/โฆ. A comprehensive ๐๐ฒ๐ฐ๐ต๐ป๐ถ๐ฐ๐ฎ๐น ๐ฟ๐ฒ๐ฝ๐ผ๐ฟ๐ detailing our methods and findings is also on the way. Stay tuned!
Oct 10, 2026 ยท 2:36 PM UTC
18
7
44
35,524
@SenseTime_AI Incredible results evolving harnesses without touching model weights
1
0
1
30
@SenseTime_AI RSI loop turning failures into lessons is brilliant approach
1
0
1
47
@SenseTime_AI 27.86 over GPT 5 Astra without updating params is wild awesome work
1
0
1
39
@SenseTime_AI RSI loop without param updates is such an elegant approach โ +27.86 vs GPT-6 Astra is massive
1
0
1
29
@SenseTime_AI Learning from every failure is brilliant excited for the open source release
1
0
1
27
@SenseTime_AI Turning real-world failures into reusable lessons is a huge step toward more reliable and adaptable robotics.
1
0
1
20
@SenseTime_AI This is incredible work. Turning failures into lessons with RSI is such a smart loop. Canโt wait for the open source drop
1
0
1
24
@SenseTime_AI ๅบๅฎๆจกๅๆ้ๆน่ฟHarnessๅพๅผๅพๅ
ณๆณจใๆๅพ
ๆฅๅๆๅผ่งๅใๅ้ฆๅๅค็นEEF้ขๆต็ๅ่ชๅข็๏ผๅนถๆซ้ฒๆฏ่ฝฎๆน่ฟ็จไบๅคๅฐไปปๅกๅ้ฆใๆจกๅ่ฐ็จๅ่ฎก็ฎๆๆฌใๅฐคๅ
ถๆณ็ๅ้ๆนๅจๅจๆช่งไปปๅกไธ็ไฟ็็๏ผๅจๅไธๆนไปปๅก่ถๆน่ถๅฅฝ๏ผไธ่ทจไปปๅก็ๆญฃๆณๅ๏ผๆฏไธค็งไธๅ็่ฟๅฑใ
1
0
1
59
@SenseTime_AI ๅคฑ่ดฅๅ็ๆน่ฟ่ฝ็ปงๆฟไธๆฅ๏ผๆขไธชไปปๅก่ฟ็ฎก็จๅ๏ผ
0
0
1
39
