General
Refinement vs. Replanning: Inference Compute Allocation in Robot Policies
arXiv cs.RO0 views
arXiv:2611.19034v1 Announce Type: new Abstract: A frozen chunked robot policy can spend deployment-time inference compute in two coupled ways: more work within each policy call to refine a predicted action chunk, or more frequent policy calls to replan from fresh observations. We study this allocation without updating model weights. Our evaluation procedure reports success with both nominal compute rate and complete attempt cost, selects comparisons on discovery tasks, and tests prespecified comparisons on held-out tasks. Comparisons added after inspecting the held-out results only summarize observed patterns. We demonstrate the procedure with one frozen SmolVLA checkpoint in synchronous LIBERO simulation, using denoising count as the refinement control and fixing chunk length at K=50. Arithmetic and latency characterize cost differently: at d=10, 82.1% of counted arithmetic is independent of denoising, whereas a descriptive affine fit to fixed-input host time assigns 17.7% to its intercept. At h=2, d=2 succeeds on 139/160 attempts versus 122/160 for d=5 (+10.6 percentage points; task-conditional 95% interval: +4.4 to +16.9), while using 5.9% less arithmetic per control step. Executing h=5 rather than h=2 at d=2 yields 138/160 versus 139/160 successes and 60.7% lower arithmetic per attempt; equivalence is not established. The case study shows why deployment adaptation must evaluate generation effort and feedback frequency jointly; it does not define a universal setting.
Source:https://arxiv.org/abs/2611.19034PreviousRobot Demonstration Videos Should Include Transparency DisclosuresNextCertify, Then Select: Viability-Aware Runtime Steering of Frozen Diffusion Policies
Related posts
Why Fibonacci?
0Hacker News
Joint Synchronization and Radar Parameter Estimation for OFDM-based DISAC Systems
0arXiv eess.SP
Movable Antenna for Anti-Jamming Transmission: Continuous and Discrete Positioning Designs
0arXiv eess.SP
From High-Level Communication Requirements to KPIs via Conformal Signal Temporal Logic Learning
0arXiv eess.SP
Least ln-cosh-based adaptive filtering algorithm for graph signal processing
0arXiv eess.SP
Asymptotics for Model Selection in Probabilistic Principal Component Analysis
0arXiv stat.TH