Searcharxiv⌕ Search

arXiv subjects

Kangjie Huang

Publications and source records attributed to Kangjie Huang.

2 recordsLinked to original sources

Seeing What Still Holds: How Waiting Control Changes Cooperative Recovery

Changing the holding reference during a fixed-duration wait can reverse the ranking of visual recovery methods. We introduce AuditEmbodied, a paired execution and trajectory evaluation framework that independently varies holding references and waiting duration while keeping recovery executors frozen. A 960-execution factorial on 20 initialization seeds compares CAD-feedback and landmark-based recovery from identical post-wait physical states within each holding and waiting condition. Protected completion requires successful recovery under the declared execution constraints. At a 1 s wait with registered wrist references, their protected completion counts change from 6/40 versus 19/40 under registered holding to 34/40 versus 22/40 under position-only holding. Trajectory analysis localizes the main separation to grasp formation: small observation-time target residuals precede large tracking errors. Joint diagnostics and controlled interventions characterize persistent tracking constraints. On fresh seeds, gradual position reanchoring improves CAD recovery's long-wait completion by 42.5 percentage points and reduces mean complete recovery time by 6.06 s relative to registered holding. Additional evaluations with pretrained policies, handover, and a mass-matched geometry characterize holding-reference sensitivity across executors and physical regimes. These findings establish holding reference as an explicit evaluation condition for comparing visual recovery performance.

cs.AI↗

AuditCoder: Process-Auditable Code Generation through Responsibility-Preserving Task Graphs

Plans can guide code generation without record?ing which code implements each subproblem, leaving reviewers to reconstruct these links dur?ing handoff and maintenance. We present AU?DITCODER, a multi-agent system for process?auditable code generation that jointly con?structs a program and its responsibility record. A contract-annotated task graph assigns sta?ble IDs before implementation and preserves code ownership through generation and assem?bly. The same IDs connect each planning commitment to its code, versioned check results, and repair history, supporting human inspec?tion and automated checks of record complete?ness and repair decisions. When failure evidence supports a node or branch, bounded repair declares an editable scope, records any expansion, and leaves code outside the final scope unchanged. The system abstains when no boundary is supported. On a 200-task APPS sample and ClassEval, AUDITCODER remains competitive with CodeSim across two models under a one-round feedback limit. The DeepSeek-based system also remains compet?itive on 22 exploratory cross-domain tasks. An audit of the DeepSeek APPS run finds that, averaged across tasks, 97.25% of final leaf records retain nonempty code, contract, provenance, and check fields. All 44 repair attempts follow the recorded accept/rollback policy, and 30 pass held-out tests, with scopes averaging 22% of graph nodes. Anonymous repository: https://anonymous.4open.science/r/AuditCoder/.

cs.SE↗