Qubit-to-Motion: Traceable Action Provenance in Hybrid QRL for Robotic Bin-Packing Sequencing
Qubit-to-Motion: Traceable Action Provenance in Hybrid QRL for Robotic Bin-Packing Sequencing
This research introduces Qubit-to-Motion, an evidence-centered framework that links quantum circuit outputs to constrained robotic actions in bin-packing tasks. The system uses four Pauli-Z readouts as input to actor and critic networks, enforces admissible object selection via a support mask, and records each decision with step indices and SHA-256 fingerprints tied to deterministic waypoints and MuJoCo contact replays. Evaluated across five 30-object scene variants, all 150 placements met support, containment, and settling criteria. A fixed-state zero-readout intervention altered 90.0% of decisions, confirming dependence on the joint circuit representation. Mean macro travel was 34.537 m, inference took 8.60 ms per route, and replay achieved a mean final position error of 8.764 μm with a maximum interpenetration of 0.356 mm. The implementation is available via an anonymous repository.
The study presents a hybrid quantum reinforcement learning framework for robotic bin-packing, demonstrating traceable action provenance through deterministic evaluations on simulated scenes.
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.
- Environment
- SimulationOpen source S3
Reported numbers
mean macro travel distance
34.537
View original evidence
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4inference time per route
8.6
View original evidence
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4mean final position error
8.764
View original evidence
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4number of scene variants
5
View original evidence
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4number of placements
150
View original evidence
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4number of Pauli-Z readouts
4
View original evidence
d sequencing and simulated robotic placement. Four Pauli-$Z$ readouts provide the sole learned input to the actor and critic, while a support mask enforces admissible object selection. Each decision is recorded and linked via step indices and SHA-256 fingerprints to a sealed route, deterministic Cartesian and inverse-kinematics waypoints, prescribed joint motion, and MuJoCo contact replay. A frozen checkpoint is eval
Open source S3
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
d sequencing and simulated robotic placement. Four Pauli-$Z$ readouts provide the sole learned input to the actor and critic, while a support mask enforces admissible object selection. Each decision is recorded and linked via step indices and SHA-256 fingerprints to a sealed route, deterministic Cartesian and inverse-kinematics waypoints, prescribed joint motion, and MuJoCo contact replay. A frozen checkpoint is eval
Open source S3
uated across five deterministic 30-object scene variants. All 150 placements satisfy support, containment, and settling criteria. A fixed-state zero-readout intervention changes $90.0\pm0.0\%$ of decisions, demonstrating dependence on the joint circuit representation. Mean macro travel is $34.537\pm0.206$~m, inference requires $8.60\pm1.27$~ms per route, and replay achieves $8.764\pm0.220~\mu$m mean final position er
Open source S4
ror with 0.356~mm maximum interpenetration. These results establish reproducible decision-to-contact traceability. The implementation is available at \href{https://anonymous.4open.science/r/Qubit-to-Motion-Quantum-Reinforcement-Learning-F7DC}{\textcolor{linkpink}{the anonymous repository}}.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org