analiza aiqbench
This paper applies learned adaptation to robot navigation: a Transformer estimates how reactive nearby pedestrians are, and that estimate gates LoRA modules on top of a frozen conservative RL policy. The reported gains (16.7% shorter paths than the conservative backbone, 14.88 points higher success rate than the reactive-trained policy) are simulation results, so real-world transfer and robustness to estimator error remain open questions. For ML readers, the practical takeaway is a modular residual-plus-gating pattern for adapting safety-first policies to heterogeneous human behavior.
Learning When to Adapt: Reactivity-Conditioned Residual Adaptation under Human-Robot Distribution Shift
arXiv:2610.25419v1 Announce Type: new Abstract: Pedestrians differ in whether they respond to a robot, yet many RL policies assume all are non-reactive to prioritize safety. Policies trained without conditioning on local pedestrians' reactivity may therefore be unnecessarily conservative or rely on cooperation that does not occur. We propose RCRA, an online adaptation framework that modulates a learned residual around a frozen conservative navigation policy according to estimated pedestrian reactivity. A Transformer-based estimator infers latent pedestrian reactivity from motion histories and robot context, providing a more accurate signal for navigation adaptation and improving scenario-averaged simulation accuracy by $12.62$ percentage points over the Bayesian baseline, which assumes known pedestrian goals. These estimates are spatially aggregated into an adaptation scale that gates LoRA modules, enabling cooperative behavior near reactive pedestrians and conservative behavior near non-reactive pedestrians. Among baselines achieving SR above $90\%$ in every scenario, RCRA reduces scenario-averaged path length by $16.7\%$ relative to the conservative backbone and by $7.3\%$ relative to randomized-training baselines. It improves mean SR by $14.88$ percentage points over the reactive-trained policy.
Abstract:Pedestrians differ in whether they respond to a robot, yet many RL policies assume all are non-reactive to prioritize safety. Policies trained without conditioning on local pedestrians' reactivity may therefore be unnecessarily conservative or rely on cooperation that does not occur. We propose RCRA, an online adaptation framework that modulates a learned residual around a frozen conservative navigation policy according to estimated pedestrian reactivity. A Transformer-based estimator infers latent pedestrian reactivity from motion histories and robot context, providing a more accurate signal for navigation adaptation and improving scenario-averaged simulation accuracy by $12.62$ percentage points over the Bayesian baseline, which assumes known pedestrian goals. These estimates are spatially aggregated into an adaptation scale that gates LoRA modules, enabling cooperative behavior near reactive pedestrians and conservative behavior near non-reactive pedestrians. Among baselines achieving SR above $90\%$ in every scenario, RCRA reduces scenario-averaged path length by $16.7\%$ relative to the conservative backbone and by $7.3\%$ relative to randomized-training baselines. It improves mean SR by $14.88$ percentage points over the reactive-trained policy.
Robotics (cs.RO)
arXiv:2610.25419 [cs.RO]
(or arXiv:2610.25419v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.25419 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.
Text integral sub licența deschisă a publicației · 259 cuvinte