Robotics · 2026-10-30
LOGIV: Graph-Verified Execution with Causal Local Repair for Long-Horizon Robot Tasks
Enqiao Lu, Yuxiang Xie, Xingrui Yu, Yuyang Wang, Yipeng Gu, Zhenglin Wan, Pengfei Zhou, Wangbo Zhao, Yang You, Yuxiang Gao, Ivor Tsang
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
LOGIV is a graph-engineering execution layer for long-horizon robot tasks. It compiles a validated plan into a task DAG with condition-support (causal) and resource-ordering edges, checks physical validity online during execution, and uses dependency tracing to constrain a causal local repair after failures—without adding trainable components on top of existing VLA/WAM policies.
Why it matters
For long-horizon robotics, an action’s completion does not guarantee that its effects remain valid for later steps under disturbances. LOGIV’s explicit task-dependency graph and continuous verification aim to update the plan safely by restoring only affected support while retaining an unfinished continuation when possible, then symbolically validating the repaired continuation before resuming.
Method
- Compile a validated PDDL plan into a DAG where causal edges capture PDDL condition support and resource edges capture ordering/interference; execution order is checked online.
- During execution, use VLM-extracted facts plus robot state/actuator feedback to gate preconditions and verify both expected effects and whether support from START/completed actions still holds for pending nodes.
- After failure, use dependency tracing to select a limited replacement scope (constrained by retained unfinished actions), build a repair fragment, validate the full continuation with VAL, then rebuild the graph and resume.
Limitation
Repair terminates when no further expansion is possible or the budget is exhausted, without guarantees of minimality or completeness.
Abstract (from arXiv)
Agentic robots must turn a high-level goal into actions while keeping the physical conditions needed for future steps valid. A completed action does not ensure that its effects will persist: a later disturbance can invalidate support that the remaining plan still assumes. We present LOGIV, a graph engineering framework for agentic robot execution that uses an explicit task graph to connect symbolic planning, observation-based verification, and embodied action. LOGIV compiles a validated plan into condition-support and resource-ordering edges, checks their physical validity during execution, and uses dependency tracing to determine what must be restored after a failure. Repair is constrained by retained unfinished actions, and the combined continuation is symbolically validated before execution resumes. The framework operates above existing Vision-Language-Action (VLA) and World Action Model (WAM) policies without adding trainable components. Across 20 dual-arm tasks from RoboTwin~V2 and BiCoord with three execution backbones, LOGIV improves mean task success without injected disturbances by 10.67 and 15.33 percentage points, respectively. Under predefined perturbations, it achieves higher task completion and condition restoration than three adapted recovery baselines for every backbone on both benchmarks. On three real-robot tasks, successful trials increase from 12/30 to 17/30. These results support explicit, continuously checked task dependencies as a practical interface for reliable agentic robot execution.
Related papers
- Towards Agentic Robots: Model, Data, Environment, and Harness
- RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments
- SEAR: Self-Evolving Agentic Robotics
- AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
- Explore, Execute, Evolve: A Skill Acquisition and Reuse Loop for Embodied Agents
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First announced by arXiv: 2026-10-30 · Indexed by Pub-AI: 2026-10-30 UTC
LOGIV: Graph-Verified Execution with Causal Local Repair for Long-Horizon Robot Tasks
First announced by arXiv: 2026-10-30 · Original record ↗
- arxiv
- 2610.40016
Explicit source versions (1)
- v1 · Date not supplied
- Preprint available
Source date: 2026-10-30 · First detected: 2026-10-30 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · RSS guid=oai:arXiv.org:2610.40016v1; arxiv:announce_type=new; item pubDate (official metadata)
Abstract · arxiv
Agentic robots must turn a high-level goal into actions while keeping the physical conditions needed for future steps valid. A completed action does not ensure that its effects will persist: a later disturbance can invalidate support that the remaining plan still assumes. We present LOGIV, a graph engineering framework for agentic robot execution that uses an explicit task graph to connect symbolic planning, observation-based verification, and embodied action. LOGIV compiles a validated plan into condition-support and resource-ordering edges, checks their physical validity during execution, and uses dependency tracing to determine what must be restored after a failure. Repair is constrained by retained unfinished actions, and the combined continuation is symbolically validated before execution resumes. The framework operates above existing Vision-Language-Action (VLA) and World Action Model (WAM) policies without adding trainable components. Across 20 dual-arm tasks from RoboTwin~V2 and BiCoord with three execution backbones, LOGIV improves mean task success without injected disturbances by 10.67 and 15.33 percentage points, respectively. Under predefined perturbations, it achieves higher task completion and condition restoration than three adapted recovery baselines for every backbone on both benchmarks. On three real-robot tasks, successful trials increase from 12/30 to 17/30. These results support explicit, continuously checked task dependencies as a practical interface for reliable agentic robot execution.