aiqbench analysis
This research introduces TISED, an analytical framework revealing counterintuitive trade-offs between inference speed and task success in embodied AI systems like robots. It demonstrates that standard efficiency techniques such as quantization can paradoxically degrade performance on static tasks while potentially enhancing it on dynamic ones due to unique closed-loop environmental interactions. These findings are critical for optimizing real-world robotic deployments where end-to-end completion time and reliability matter more than isolated inference latency.
The Speedup Paradox: Rethinking Inference Speed-Quality Trade-off in Embodied Tasks
arXiv:2606.28529v3 Announce Type: replace-cross Abstract: Embodied foundation models have recently been widely used to improve robot generalization and task success rates. Previous works apply lossy efficient-inference techniques such as quantization, pruning, and asynchronous inference, accepting small action quality degradation in exchange for lower per-inference computation cost and inter-action latency. However, unlike traditional static ML tasks, embodied tasks involve repeated interaction with the environment, and task-level performance is determined not only by inference cost, but also by closed-loop effects unique to embodied execution, which remain insufficiently characterized in current efficient-inference studies. In this work, we propose TISED (\underline{T}ask-level \underline{I}nference \underline{S}peedup \underline{E}ffect \underline{D}ecomposition), an analytical framework that unifies diverse lossy lightweighting inference optimization techniques and decomposes their effects on static and dynamic tasks; using it, we uncover several paradoxical effects: (1) on \textit{static tasks}, optimization can sometimes lengthen end-to-end per-task completion time even as per-chunk time drops; (2) on \textit{dynamic tasks}, moderate lossy lightweighting optimization can raise task success rate even above the baseline; and (3) the monotonicity and sweet-spot location of both effects can shift with hardware configuration. Together, our findings provide a new perspective on adapting inference optimization techniques to embodied tasks. Source code: https://github.com/thu-ee-acts-lab/robotics-the-speedup-paradox.
Abstract:Embodied foundation models have recently been widely used to improve robot generalization and task success rates. Previous works apply lossy efficient-inference techniques such as quantization, pruning, and asynchronous inference, accepting small action quality degradation in exchange for lower per-inference computation cost and inter-action latency. However, unlike traditional static ML tasks, embodied tasks involve repeated interaction with the environment, and task-level performance is determined not only by inference cost, but also by closed-loop effects unique to embodied execution, which remain insufficiently characterized in current efficient-inference studies. In this work, we propose TISED (\underline{T}ask-level \underline{I}nference \underline{S}peedup \underline{E}ffect \underline{D}ecomposition), an analytical framework that unifies diverse lossy lightweighting inference optimization techniques and decomposes their effects on static and dynamic tasks; using it, we uncover several paradoxical effects: (1) on \textit{static tasks}, optimization can sometimes lengthen end-to-end per-task completion time even as per-chunk time drops; (2) on \textit{dynamic tasks}, moderate lossy lightweighting optimization can raise task success rate even above the baseline; and (3) the monotonicity and sweet-spot location of both effects can shift with hardware configuration. Together, our findings provide a new perspective on adapting inference optimization techniques to embodied tasks. Source code: this https URL.
Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2606.28529 [cs.RO]
(or arXiv:2606.28529v3 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.28529 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 292 words
Related coverage
Can we change an AI model’s architecture without retraining it from scratch?
Amazon Science · Oct 12, 2026
DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models
arXiv · Oct 6, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
[AINews] Reflection Beam - 501B-A23B American Open Model
Latent Space · Oct 6, 2026