ChunkAct: Action Chunking for Agentic Robotics
ChunkAct: Action Chunking for Agentic Robotics
ChunkAct is a novel approach to agentic robotics that introduces action chunking to reduce computational overhead. By grouping primitives into autonomous chunks, it improves task success on LIBERO-Pro while requiring fewer LLM calls, input tokens, and output tokens compared to existing methods.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
task success
5
View original evidence
ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics
Open source S4LLM calls
2.9
View original evidence
ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls
Open source S4input tokens
4.4
View original evidence
ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls, 4.4x fewer input
Open source S4LLM calls
2.9
View original evidence
ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls, 4.4x fewer input
Open source S4
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ine, ChunkAct compiles the memory into these chunks, so execution is memory-free at test time and the planner is reserved for decisions that require reasoning. LLM computation therefore scales with the number of decisions rather than the number of primitives. On LIBERO-Pro, ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls, 4.4x fewer input
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org