Searcharxiv⌕ Search

arXiv subjects

Shuning Zhang

Publications and source records attributed to Shuning Zhang.

At least 19 recordsLinked to original sources

AgenticJevNav: Agentic Object Search and Navigation via Event-Driven Verification

Agentic navigation must maintain goals and adapt actions from incomplete evidence despite execution failures and delayed inference. We present AgenticJevNav, which links goal-directed planning, evidence memory, event-triggered verification, and bounded execution. The Agent plans from RGB, odometry, memory, and feedback, retaining goal ownership while authorizing a bounded action set. Jev reviews structured evidence after blocked/failed execution or two consecutive low-progress segments. Version, freshness, and once-only checks govern decision application. A single executor and independent RGB reflex constrain motion. Goal revisions revoke obsolete requests and actions before replanning. Across six simulated task families, AgenticJevNav achieves 73.33% SR over 60 trials, 21.67 percentage points above the next-best Agent-only aggregate control under differing historical protocols. A real-world GENISOM L1 quadruped deployment achieves 70.0% success over 30 trials, including bounded search reporting. Together, the simulation and physical-robot results demonstrate the effectiveness of the proposed architecture, while remaining failures highlight opportunities to improve grounding, approach planning, and runtime.

cs.RO↗

From LLM Hallucinations to Inspiration in Creative Work

Large language models (LLMs) can introduce unexpected ideas through hallucinations, yet little is known about when creative practitioners find these errors inspiring or how these judgments shape interactions. We conducted a retrospective qualitative study with 20 experienced practitioners in research, design, and social media content creation, thematically analyzing 41 hallucination cases and 20 semi-structured interviews. Our framework links inspiration judgments of hallucinations based on goal relevance, task stage, content adaptability, and type-specific factors to behaviors across not-inspired, inspired-by, and inspired-to states. We characterize downstream interaction consequences, showing how hallucination-inspired experiences vary across creative stages and reshape role-differentiated trust, and propose these consequences as a dimension of hallucination evaluation. Finally, we derive design recommendations for stage-sensitive control of divergence and adherence, user-defined constraint priorities, differentiated hallucination labels, and salvage support for preserving useful ideas while correcting errors. These contributions deepen HCI understanding of hallucinations and responsible LLM design in creative work.

cs.HC↗

Mitigating Agentic Deception via Evaluation-Side Partially Observable Games

As large language models become more capable agents, they can exhibit deceptive behaviors, cheating human for higher rewards while appearing helpful. We attribute this failure to evaluation-side partial observability, where models are evaluated mainly from final outcome, without checking the faithfulness of underlying reasoning, tool use, and environmental effects. Prior mitigations address this partial observability by predefined evaluator latent belief space or one-shot monitoring, which do not extend to free-form language space and vulnerable to stronger deceptions that evade simple monitoring. In this work, we formulate deception mitigation as an Evaluation-Side Partially Observable Extensive Form Game (e-PoG), a max-min framework in which an agent maximizes reward, a monitor minimizes the agent's payoff by exposing deception from the full interaction trajectory, and a fixed verifier validates the monitor's claims and converts them into penalties. We define the solution concept as a Subgame-Perfect Oversight Equilibrium (SPOE) between agent and monitor, and prove its existence in finite games. With a perfect verifier, every SPOE is faithful; with an imperfect verifier, the guarantee persists if verifier errors do not reverse the ranking of monitor actions. To achieve faithful SPOE, we derive a two-timescale policy gradient algorithm and enhance the pipeline empirically with NLI verification and deterministic checks. Experiments with Gemma3-4B-IT and Qwen3-4B/8B on Agentic Role-Playing, SearchQA, and WebShop show that e-PoG mitigates multiple emergent deception modes, outperforming the strongest baselines by 26.0% in deception mitigation and by up to 39.8% in task success. These gains persist under weak-to-strong supervision. Code is available at https://github.com/Toyoid/deceptive-agent.

cs.LG↗

ForeAct: Geometry-Guided Action Learning from World Representations

Robotic manipulation requires representations that capture both scene geometry and its evolution during interaction. Existing approaches have largely treated these two aspects separately. World action models capture such dynamics by coupling action prediction with future video generation, but their reliance on iterative diffusion or rectified-flow inference leads to high computational cost, while their representations lack explicit spatial structure. Explicitly modeling scene geometry typically requires auxiliary modules that increase both parameter count and computational cost. Conversely, geometry-centric policies primarily recover observed spatial structure and often lack explicit supervision for anticipating future geometric changes during interaction. To address these limitations, we introduce ForeAct, which repurposes a pretrained video world model as a single-pass representation backbone for low-latency control. ForeAct directly processes the clean latent of the current observation in one forward pass, eliminating future-frame synthesis and iterative denoising while retaining the backbone's pretrained spatiotemporal priors. Rather than introducing a separate geometry backbone, ForeAct uses learnable tokens supervised by next-frame relative depth to shape the shared video representation through bidirectional self-attention. This compact interface turns the spatial knowledge already encoded in the backbone into a future-aware representation for action generation, with little parameter overhead. The lightweight geometry readout is used only during training and removed at deployment, enabling low-latency, geometry-aware control without depth input or an external geometry backbone. Without additional large-scale pretraining, ForeAct achieves 96.6% average success on LIBERO, 63.5% on SimplerEnv-Bridge, and a 74.4% success rate across 90 real-world rollouts.

cs.CV↗

Kimodo-Policy: Learning Kinematic Action Models for Humanoid Loco-Manipulation

Humanoid loco-manipulation requires continuous coordination of high-dimensional whole-body motion to realize task objectives. While vision-language-action models and world action models transfer semantic and visual-dynamics priors to robot control, these priors do not necessarily capture the structure of robot motion. We introduce Kinematic Action Models (KAMs), a paradigm that treats whole-body motion as a direct pretraining target and uses a pretrained motion generator as the primary action-generation backbone. KAMs ground learned temporal and cross-body coordination priors through task conditioning, visual observations, and execution feedback. We present Kimodo-Policy as an initial instantiation, combining a frozen Kimodo text-to-motion backbone with lightweight ControlNet-based visual conditioning. A root-body-hand hierarchy coordinates locomotion and manipulation, while motion-history conditioning and repeated action-chunk prediction bridge motion synthesis and closed-loop control. Following large-scale loco-manipulation pretraining, the policy achieves the highest overall success among compared policies on multi-task HumanoidArena (72.14%), competitive performance on SIMPLE, and 60% mean success across eight real-world tasks, with 678M parameters (92M trainable). Ablations support the contribution of the pretrained motion backbone. Together, these results provide initial evidence for motion generation as a complementary foundation for humanoid action learning.

cs.RO↗

ConMem: Counterfactual Multi-Scenario Probing for Resolving LLM Memory Misalignment

Long-term memory is critical for autonomous Large Language Model (LLM) agents executing multi-session, long-horizon tasks. However, existing memories often inappropriately store, retain, or retrieve over-generalized, misattributed, or privacy-sensitive information, a problem we term memory misalignment. To study this problem, we construct \textbf{MemAlign}, a benchmark comprising 1,079 curated memory cases grounded in synthetic, validated multi-session conversational contexts covering diverse misalignment taxonomies. We then propose ConMem, an algorithm that reframes retention decisions as prospective counterfactual causal inference. \textbf{ConMem} synthesizes counterfactual interaction scenarios to prospectively audit downstream intervention effects across three aspects: relevant task execution, irrelevant invariance checks, and risk-sensitive boundary shifts. Experiments show that ConMem outperforms other methods, achieving 75.16\% accuracy, 64.15\% balanced accuracy, and a Macro-F1 of 65.31\%, effectively mitigating over-retention biases without sacrificing personalization utility.

cs.AI↗

From Neglect to Resignation: Understanding the Evolution of Mobile Privacy

This decade-long longitudinal qualitative study follows 31 smartphone users in China across four interview waves (2014-2023) to examine how mobile privacy understanding evolves alongside shifting computing infrastructures. Findings reveal three stages shaped by distinct interactional mechanisms: early functional acceptance of permission-based data access, growing vigilance triggered by behavioral tracking feedback loops, and eventual resignation as AI-driven inference made ecosystem-level profiling feel inescapable. Rather than a stable attitude, digital resignation emerges through cumulative infrastructure exposure, as each new system layer retroactively reinterpreted earlier permissions and exhausted users' capacity for self-management. The study challenges permission-centric design assumptions and proposes inference visibility, context isolation, and automated privacy agency as system-level responses.

cs.HC↗

MisleadQA: Diagnosing LLM Failures under Misleading Context

Large language models (LLMs) are widely used to answer questions from retrieved or user-provided context. However, provided context can be factually true yet out-of-scope, describing a different entity, time, or location from the one the question targets, and still lead models to a wrong answer. Prior work has mainly studied factually false or irrelevant context. It remains unclear whether such out-of-scope context misleads models as much as false context, and whether adding gold context prevents models from being misled by either. We introduce **MisleadQA**, a diagnostic dataset of multiple-choice questions with an abstention option. Each question pairs gold context (G) with true but out-of-scope context (S) and false but in-scope context (F) supporting the same wrong answer. We evaluate six open-weight LLMs. On questions they answer correctly without context, providing S alone leads them to switch to this wrong answer in 29.3% of cases, and providing F alone does so in 75.4%. Adding G nearly removes the effect of S, but F still misleads models in 21.0% of cases, especially when F comes first. A context-faithful instruction reduces these errors mainly through abstention and lowers accuracy even with G. These findings suggest that evaluations should check whether context applies to the question, not only whether it is true, and should report correct, misled, and abstention rates together. Data and code are available at [https://anonymous.4open.science/r/MisleadQA](https://anonymous.4open.science/r/MisleadQA).

cs.CL↗

Covert Digital Semantic Communication With Movable Antennas: Constellation Shaping, Spatial Optimization, and Semantic Mapping

Digital semantic communication (SemCom) is emerging as a bandwidth-efficient paradigm for next-generation wireless networks by delivering the meaning of a source over finite constellations. However, the broadcast nature of wireless channels exposes transmissions to adversarial wardens, while discrete symbols deviate from the Gaussian signaling assumed by conventional covert communication. In this paper, we propose a covert digital SemCom system assisted by movable antennas (MAs), where the constellation, the antenna positions, and the semantic transceiver are jointly designed to conceal the transmission while preserving the semantic fidelity. Specifically, the maximum covert rate against the optimal detector with unknown noise level is derived in closed-form, which reveals that the legitimate signal-to-noise ratio (SNR) scales with the moment-matching order of the constellation and the spatial gain provided by the MA array. A Gauss-Laguerre ring constellation is then designed to raise the moment-matching order under the peak power constraints, and a covert symbol mapping with rank-to-ring assignment and scrambling is developed to carry the semantics in the ordering of the latent magnitudes. Guided by the derived semantic distortion bound, the semantic encoder and decoder are trained end-to-end through an invertible flow over the designed constellation and MA array. Simulation results demonstrate that the proposed scheme supports a higher achievable covert rate and improves the reconstruction quality by 24% compared to state-of-the-art SemCom schemes.

eess.SP↗

Cooperative Multi-Agent Vision-Language-Action Models via Reinforced Fine Tuning

We study reinforcement learning (RL) methods for cooperative multi-agent Vision-Language-Action (VLA) models. This problem is challenging because VLAs are pretrained on large-scale single-agent data and therefore lack the fine-grained coordination skills required for inter-robot collaboration. Supervised fine-tuning (SFT) on multi-robot demonstrations partially bridges this gap, but its performance is bounded by the demonstration data and cannot improve from its own experience. We present a three-stage reinforced fine-tuning (RFT) pipeline for multi-agent VLAs. First, initialization-aware data collection sweeps over initial configurations and invokes human demonstrations only when the pretrained VLA repeatedly fails, yielding robustness to initialization shift with reduced human cost. Second, offline credit-filtered tuning assigns credit to individual agents and fine-tunes on per-agent trajectories with positive advantage rather than on entire joint rollouts. Third, we find existing online RL for VLAs are less effective for hard multi-agent tasks, which we attribute to noisy co-exploration and unstable updates. We instead use online latent-space fine tuning, which freeze the VLA and perform RL in its latent noise space. We evaluate our multi-agent VLA with both $π_0$ and $π_{0.5}$ backbones across 11 tasks in RoboTwin, RoboFactory and real-world manipulation with two Franka robots. Our multi-agent VLA improves the average success rate by $+23.1\%$, $+16.4\%$, and $+44\%$ on RoboTwin, RoboFactory, and real-world tasks, respectively. Code available at https://anonymous.4open.science/r/mavla_rft-2BC0/.

cs.RO↗

"Black Mirror?": Public Sensemaking of AI-Powered Lifelogging

AI-powered lifelogging wearables are emerging as a new class of consumer devices that transform everyday experience into searchable, AI-curated memory archives. We study early public sensemaking around these systems at the moment of their market entry, using the Looki L1 as an empirical lens. Analysing large-scale Chinese-language and English-language social media discourse (N = 5,053 comments), we combine topic clustering with inductive thematic analysis to examine how users interpret the social, moral, and political implications of AI-mediated memory. Across contexts, users reference dystopian surveillance imaginaries, express privacy resignation and bystander concerns, and debate assistive value alongside consumer logics. English-language comments more often framed these devices through interpersonal power, evidentiary use, and hacking anxieties, while Chinese-language comments more often foregrounded labour exploitation, governance surveillance, and technological inevitability.

cs.HC↗

Characterizing Memory Misalignment in Human-LLM Interaction From User Perspectives

While memory enhances personalization in LLM-based conversational agents, it suffers from memory misalignment, where memories violate user expectations. We present a mixed-methods investigation to characterize and mitigate memory misalignment from user perspectives. First, we collected data from memory usage (N=28, 457 entries) and diary study (N=32, 304 reports), which yielded a taxonomy spanning 14 misalignment types across memory intake, storage and management, retrieval and interpretation stages. Second, four co-design workshops with 12 experienced HCI researchers derived a design space to tackle memory misalignment issues, consisting of 12 candidate interaction strategies structured across interaction form, placement and intrusiveness dimensions. Finally, a speed dating with 121 users reveals preference heterogeneity, where users prioritize proactive controls over cognitively demanding causal graph inspections or passive audit logs. Synthesizing these findings, we highlight the tension between supervisory agency and interaction overhead, and advocate for friction-aware memories that balance user oversight with conversation smoothness.

cs.HC↗

eTracer: Polarity-Aware Evidence Grounding

Evidence grounding makes generated content verifiable by linking statements to their sources. However, effective verifiability requires two properties that existing methods fail to jointly provide: verification granularity, the ability to localize fine-grained evidence for each statement, and verification completeness, the ability to capture both support and contradiction. Existing methods either operate at granularities too coarse for rapid inspection or are support-only. We formalize polarity-aware evidence grounding: linking each response sentence to its contextual evidence with a supporting, contradicting, or irrelevant label. We construct eTracer-Bench, a human-annotated benchmark with 7,905 supporting and contradicting sentence pairs across seven domains, covering both explicit and implicit contradictions. We further propose eTracer, a post-hoc framework whose context-aware signed scorer recovers the full evidence map in a single forward pass per response, with localization and polarity emerging jointly from one mechanism. eTracer achieves the best overall Citation F1 of 82.39, driven by a Contradict F1 of 89.20. A pre-registered within-subjects user study (n = 30) shows that polarity-aware grounding nearly doubles contradiction detection over the document-level baseline (27.9% to 51.9%, Cohen's d = 0.82, p = 0.0003), and suggests a false-completeness bias under which support-only grounding fails to improve over this baseline (23.4% vs. 27.9%). Dataset and code: https://github.com/chubohao/eTracer.

cs.CL↗

Locally Sound, Globally Insufficient: The Local-Global Gap in Multi-Hop Reasoning

Reliable multi-hop reasoning requires more than locally supported steps: a trace can be sound at every reasoning step yet still fail to answer the question as a whole. We call this failure regime the local-global gap (LGG), in which the trace is locally sound yet globally insufficient. Local soundness requires each step to be supported by the available evidence and preceding steps, whereas global sufficiency requires the reasoning trace to align with the question and establish the submitted answer. In a human-adjudicated diagnostic of 2,598 responses across three multi-hop QA benchmarks and three models, we find that the LGG occurs in every benchmark-model combination and accounts for nearly half of globally insufficient responses overall. However, conventional faithfulness verifiers that check claims against the evidence largely miss these failures: at thresholds retaining at least 95% of reliable traces, recall for LGG cases is substantially lower than that for locally unsound traces. To address these failures, we formalize three dependencies for reliable reasoning: evidence-to-step support, question-to-trace alignment, and trace-to-answer closure. Instead of post-hoc diagnosis, we introduce E-Closure to supervise these dependencies during training, combining generation supervision on supported original and counterfactual responses with bidirectional switching constraints. Averaged over three benchmarks and three backbones, existing fine-tuning baselines improve accuracy and local soundness over the base models, but at the cost of global sufficiency. E-Closure improves both: among all fine-tuned methods, it achieves the highest average accuracy (92.8%) and trace reliability (89.0%) while yielding the lowest LGG rate (6.2%).

cs.CL↗

Alignment and Divergence between Humans and AI in Interpersonal Privacy Decisions

AI assistants increasingly mediate interpersonal communication on behalf of their primary user, but they risk violating the privacy expectations of third-party information owners. Resolving these tensions requires understanding how humans anticipate interpersonal privacy boundaries. Therefore, we conducted a dyadic study (N=76) and a matched evaluation of AI models across 18 information types and 3 recipient relationships. We found that data owners' privacy judgments are highly contextual and relationship dependent. While familiar data co-owners show meaningful alignment with owners' expectations, they significantly overestimate the need for permission. Interestingly, greater familiarity within the owner-co-owner dyad was associated with both higher disclosure acceptability and lower co-owner misalignment, whereas our exploratory four-item empathy measure was not. In contrast, AI models significantly underperform human co-owners in anticipating the data acceptability, even when provided with within-dyad examples. These findings underscore a core HCI design challenge to develop privacy-aware AI that respects multi-stakeholder information boundaries.

cs.HC↗

AI Peers Exert Social Influence on Human Dishonesty in Groups

Human dishonesty in group settings is highly susceptible to peer influence, particularly when incentivized. Although artificial intelligence (AI) evolves from passive tools into active collaborators, its impact on human moral behavior within groups remains underexplored. We addressed this gap through a two-phase randomized behavioral study (N=280 and N=360). We found AI agents exert substantial social influence comparable in magnitude to that of human peers. Specifically, participants reported more dishonestly when exposed to dishonest rather than honest normative cues. This effect is evident across injunctive, subjective, and descriptive social norms. Interestingly, the only significant adjacent behavioral change occurred when dishonest peer behavior first appeared, whereas further increases from one to four dishonest peers produced weaker and non-monotonic changes. Furthermore, participants rapidly converge on decision-making, showing modest increases in dishonest reporting through repeated exposure. These findings highlight the importance of managing the behaviors and normative signals communicated by AI group members.

cs.HC↗

Needs Your Help: Understanding Platform-Directed Rating Participation in Community Notes on X

Community-based fact-checking is promising in countering misinformation, yet its scalability is constrained by slow rating accumulation. To address this challenge, platforms such as X implement platform-directed rating mechanisms, specifically through ``Needs Your Help'' algorithmic prompts, to target unresolved notes. Using a dataset of over 220 million rating contributions -- including 1.9 million platform-directed ratings -- on X, we examine contributors' response to note prompts, notes' resolution, and raters' spillovers. We found (i) at note level, population-sampled ratings concentrate on recent notes with certain helpfulness and high disagreement. Once sampled, population-sampled rating was associated with faster and more transitions to resolved statuses. (ii) At rater level, following raters' first observed population-sampled rating, raters exhibit significant yet modest increases in daily ratings, rating pace and tag usage, while other behaviors show no change. These highlight the promise of algorithmic nudges to guide volunteer attention toward contested content, accelerating consensus while sustaining rater engagement.

cs.HC↗

Air-Ground Collaborative Vision-and-Language Navigation via Shared Bird's-Eye Maps

Air-ground collaborative Vision-and-Language Navigation (VLN) pairs an unmanned aerial vehicle (UAV) with a global bird's-eye view and an unmanned ground vehicle (UGV) with a local first-person view, yet the setting remains largely unexplored: existing training-free methods solve single-agent tasks but offer no collaboration mechanism, and a recent CARLA-Air evaluation found no stable cooperative behavior across five state-of-the-art VLA models; naive semantic communication or bidirectional coupling even degrades performance. We establish AGC-VLN (Air-Ground Collaborative VLN), the first training-free baseline for air-ground collaborative VLN. The key insight is that training-free methods decompose navigation into VLM-based semantic reasoning and deterministic geometric execution, exposing a collaboration interface: the UAV's global view, over which it renders the UGV's reported pose and the VLM-anchored target as CAR/GOAL markers with distance labels, yielding a shared bird's-eye map. From this map, the UGV acquires global spatial context its first-person view cannot provide, plans a road-following path with a frozen VLM, and executes it under closed-loop control; in parallel, the UAV runs 3D-SPF, a spatial-search upgrade of SPF that localizes the target in the downward view and flies toward it. On 100 closed-loop episodes in CARLA-Air's Town10HD scene, AGC-VLN reaches a 77.0% joint success rate, a collaboration gain of +27.0% over the weaker individual agent (the UAV, 50.0%), and exceeds the strongest published single-agent baseline (Travel UAV, 53.0%) by 24.0 points, stemming from the complementarity of the UAV's global view and the UGV's road-following execution. Project page: https://github.com/ZSN2024/AGC-VLN.

cs.RO↗