Search papers, labs, and topics across Lattice.
The paper introduces $\chi_{0}$, a resource-efficient framework designed to improve the robustness of long-horizon robotic manipulation by addressing distributional shifts between demonstration, policy learning, and execution. $\chi_{0}$ tackles these inconsistencies using model arithmetic for demonstration diversity, stage-aware advantage estimation for stable training signals, and train-deploy alignment techniques like spatio-temporal augmentation and DAgger corrections. Experiments on dual-arm garment manipulation show $\chi_{0}$ achieves a 250% improvement in success rate over $\pi_{0.5}$ while using less data and compute.
Achieve 2.5x higher success in long-horizon robotic manipulation with 90% less data and compute by explicitly aligning training and deployment distributions.
High-reliability long-horizon robotic manipulation has traditionally relied on large-scale data and compute to understand complex real-world dynamics. However, we identify that the primary bottleneck to real-world robustness is not resource scale alone, but the distributional shift among the human demonstration distribution, the inductive bias learned by the policy, and the test-time execution distribution -- a systematic inconsistency that causes compounding errors in multi-stage tasks. To mitigate these inconsistencies, we propose $\chi_{0}$, a resource-efficient framework with effective modules designated to achieve production-level robustness in robotic manipulation. Our approach builds off three technical pillars: (i) Model Arithmetic, a weight-space merging strategy that efficiently soaks up diverse distributions of different demonstrations, varying from object appearance to state variations; (ii) Stage Advantage, a stage-aware advantage estimator that provides stable, dense progress signals, overcoming the numerical instability of prior non-stage approaches; and (iii) Train-Deploy Alignment, which bridges the distribution gap via spatio-temporal augmentation, heuristic DAgger corrections, and temporal chunk-wise smoothing. $\chi_{0}$ enables two sets of dual-arm robots to collaboratively orchestrate long-horizon garment manipulation, spanning tasks from flattening, folding, to hanging different clothes. Our method exhibits high-reliability autonomy; we are able to run the system from arbitrary initial state for consecutive 24 hours non-stop. Experiments validate that $\chi_{0}$ surpasses the state-of-the-art $\pi_{0.5}$ in success rate by nearly 250%, with only 20-hour data and 8 A100 GPUs. Code, data and models will be released to facilitate the community.
