Dual-Level Visual Guidance System

• Dual-level visual guidance systems are design patterns that split visual tasks into a slow, semantic planning layer and a fast, reactive execution layer.
• They integrate high-level cues like global context or semantic embeddings with low-level signals such as pixel goals and motion cues for complex decision processes.
• Empirical studies in robotics, video analysis, and AR show improved accuracy, reduced latency, and enhanced system interpretability compared to monolithic approaches.

Searching arXiv for the cited papers and closely related work on dual-/multi-level visual guidance. A dual-level visual guidance system is a design pattern in which visual guidance is split across two interacting levels rather than handled by a single monolithic mapping. Across recent work, the term is used for systems that separate a slow, deliberative, or global layer from a fast, reactive, local, or execution layer, with the two levels coupled by structured interfaces such as visual prompts, semantic embeddings, pixel goals, floorplan patches, motion cues, or AR overlays (Wang et al., 23 Mar 2026). In this usage, the high-level layer typically specifies what and where, while the low-level layer specifies how; the result is a family of architectures for robotics, video understanding, floorplan localization, augmented reality, assistive navigation, visual generation, image restoration, and mixed-reality training (Peng et al., 24 Aug 2025).

1. Concept and defining characteristics

In robotics, the pattern is stated explicitly as a split between “a slow, deliberative ‘System 2’ that reasons in language and scene semantics, and a fast, reactive ‘System 1’ that executes precise motor control, with a visual prompting interface in between” (Wang et al., 23 Mar 2026). In video understanding, the same structure appears as “global, event-level guidance” together with “local, sub-event-level guidance,” where an LVLM supplies semantic descriptions that steer lower-level spatio-temporal processing (Peng et al., 24 Aug 2025). In floorplan localization, the term refers to “first get depth-based candidates, then use dual-level visual-geometric contrastive signals to pick the correct one,” namely position-level and orientation-level guidance (Meng et al., 5 Jan 2026).

This suggests that the phrase does not name a single algorithm. It names a recurrent systems principle: one level provides stable, semantically meaningful, or globally constraining guidance, while another level uses denser, lower-latency, or more spatially specific signals to complete perception, control, or decision making. In some systems the split is architectural and asynchronous, as in fast-slow navigation models (Zhu et al., 9 Jul 2026); in others it is representational, as in the use of semantic and structural priors in visual text generation (Luo et al., 10 Jan 2025); and in still others it is user-interface driven, as in the separation between task planning and situated AR rendering (Zhao et al., 5 Aug 2025).

The high-level layer is generally associated with instruction interpretation, task decomposition, global path selection, semantic summarization, or user-facing strategy selection. The low-level layer is generally associated with precise grounding, local motion, temporal calibration, candidate disambiguation, or visually explicit execution cues. A plausible implication is that dual-level guidance is most useful when a single stage would otherwise have to solve semantically heterogeneous subproblems at once, such as instruction parsing, grounding, spatial disambiguation, and motor generation.

2. Architectural patterns and interfaces

A common pattern is planner–controller decomposition. In "VP-VLA: Visual Prompting as an Interface for Vision-Language-Action Models" (Wang et al., 23 Mar 2026), standard VLA policies of the form $a_t = \pi_{\theta}(l, o_t)$ are replaced by a dual-system architecture in which a “System 2 Planner” decomposes the task into subtasks, identifies target objects and target locations, and generates visual prompts, while a “System 1 Controller” consumes raw observations and the prompt image to output action chunks. The controller policy becomes $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$, and the visual prompt image $I_{vp}^t$ is the System 2 $\rightarrow$ System 1 interface (Wang et al., 23 Mar 2026).

A second pattern is global-semantic and local-refinement coupling. In "Multi-Level LVLM Guidance for Untrimmed Video Action Recognition" (Peng et al., 24 Aug 2025), a Cross-Modal Guidance Branch uses an LVLM to generate a single “Global Event Prompting” narrative and multiple “Temporal Sub-event Prompting” descriptions, which are then injected into the video encoder through adaptive gating, cross-modal attention, and an event graph module. Here the interface is not an image overlay but a set of semantic embeddings and graph relations that modulate visual features at different temporal scales (Peng et al., 24 Aug 2025).

A third pattern is coarse candidate generation followed by local disambiguation. In "DisCo-FLoc: Using Dual-Level Visual-Geometric Contrasts to Disambiguate Depth-Aware Visual Floorplan Localization" (Meng et al., 5 Jan 2026), a Ray Regression Predictor produces a Depth-Aware FLoc Probabilistic Map, Top-$X$ pose candidates are selected, and a second-stage contrastive module evaluates local floorplan patches using position-level and orientation-level negatives. The two levels are fused as $$P_{\text{final}(S_i)} = (1 - w)\,P_d(S_i) + w\,P_{\text{CL}(S_i)}$$, with $w = 0.5$ by default (Meng et al., 5 Jan 2026).

A fourth pattern is slow global reasoning with fast local control. In "FSD-VLN: Fast-Slow Dual-System Modeling for Aerial Long-Horizon Vision-Language Navigation" (Zhu et al., 9 Jul 2026), the slow stream extracts stable semantic priors from a pre-trained VLM, while a Diffusion Transformer fast stream models cross-temporal action distributions; the two branches run asynchronously. In "Ground Slow, Move Fast: A Dual-System Foundation Model for Generalizable Vision-and-Language Navigation" (Wei et al., 9 Dec 2025), System 2 predicts a mid-term waypoint as a 2D pixel goal and a latent goal representation, while System 1 uses those signals plus high-frequency RGB to generate smooth continuous trajectories at much higher frequency (Wei et al., 9 Dec 2025).

In user-interface systems, the interface often becomes a guidance rendering layer. "Guided Reality: Generating Visually-Enriched AR Task Guidance with LLMs and Vision Models" (Zhao et al., 5 Aug 2025) separates a semantic level, where ChatGPT-4o generates step-wise instructions and assigns one of five visual guidance types, from a spatial level, where Gemini, SAM 2.1, ARKit, and SceneKit localize and render highlights, movement animations, hand gestures, tool usage, and timers. "JARVIS: A Just-in-Time AR Visual Instruction System for Cross-Reality Task Guidance" (Sun et al., 11 Apr 2026) similarly separates pre-task planning and in-task state verification from the visual renderer that grounds bounding boxes, shape previews, arrows, and embodiment cues into the scene (Sun et al., 11 Apr 2026).

3. Guidance representations and learning mechanisms

The interface between levels varies by domain, but the representations are consistently structured. In VP-VLA, the planner outputs a visual prompt representation $\psi_t = \{C, B\}$, where $C \in \mathbb{R}^2$ is a crosshair placed at the centroid of the target object and $B \in \mathbb{R}^4$ is a bounding box marking the placement or goal region; these are rendered on the overhead camera view to produce $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$0 (Wang et al., 23 Mar 2026). In DualVLN, the high-level planner outputs a 2D pixel goal in the current image together with a latent goal representation extracted by learnable query tokens appended to the VLM sequence (Wei et al., 9 Dec 2025). In AerialVLA, the “minimalist dual-view perception” uses a vertical mosaic $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$1 so that the front view provides critical cues for obstacle avoidance and target identification, while the down view is essential for precise ground alignment and landing maneuvers (Xu et al., 15 Mar 2026).

Training objectives likewise reflect the separation of levels. VP-VLA adds an auxiliary visual grounding loss to ensure that the controller actually uses the prompts: image coordinates are discretized into $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$2 bins, cross-entropy is used for grounding, and the total loss is an action loss plus event-triggered grounding loss with $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$3 (Wang et al., 23 Mar 2026). ECVT combines classification loss, temporal GIoU regression, a semantic consistency loss between calibrated visual features and LVLM sub-event embeddings, and a temporal calibration loss anchored to an event graph (Peng et al., 24 Aug 2025). DisCo-FLoc combines a ray regression loss for depth-aware candidate generation with a PointInfoNCE-style contrastive loss that includes both position-level negatives and orientation-level negatives (Meng et al., 5 Jan 2026).

In diffusion-based generation and restoration, the two levels can be semantic and structural rather than planner and controller. "Beyond Flat Text: Dual Self-inherited Guidance for Visual Text Generation" (Luo et al., 10 Jan 2025) defines a Semantic Rectification Branch that injects a flat-text semantic prior and a Structure Injection Branch that injects a glyph-structured prior, both by masked AdaIN and weighted fusion. "HAODiff: Human-Aware One-Step Diffusion via Dual-Prompt Guidance" (Gong et al., 26 May 2025) constructs a positive prompt from a high-quality branch and a negative prompt from residual-noise and HMB-segmentation branches, then combines the resulting predictions by $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$4 (Gong et al., 26 May 2025). This suggests that dual-level guidance need not be hierarchical in time; it can also be hierarchical in representation, with one level correcting content and another constraining structure or degradation.

4. Representative instantiations across domains

The pattern appears in a wide range of application areas.

Domain	Representative system	Two guidance levels
Vision-language-action robotics	VP-VLA (Wang et al., 23 Mar 2026)	System 2 Planner / System 1 Controller
Untrimmed video action recognition	ECVT (Peng et al., 24 Aug 2025)	Global Event Prompting / Temporal Sub-event Prompting
Floorplan localization	DisCo-FLoc (Meng et al., 5 Jan 2026)	Depth-aware candidate map / position-orientation contrastive disambiguation
AR task guidance	Guided Reality (Zhao et al., 5 Aug 2025)	Instruction & visual strategy / visual grounding & rendering
VLN and aerial navigation	FSD-VLN (Zhu et al., 9 Jul 2026), DualVLN (Wei et al., 9 Dec 2025)	slow semantic planner / fast diffusion-based controller
UAV VLA control	AerialVLA (Xu et al., 15 Mar 2026)	front-view coarse navigation / down-view precision landing

Assistive and human-centered systems extend the same principle to perception, instruction, and accessibility. "NavVI: A Telerobotic Simulation with Multimodal Feedback for Visually Impaired Navigation in Warehouse Environments" (Maimuna et al., 20 Jul 2025) has a higher-level path-planning layer and a lower-level proximity/obstacle layer, expressed across visual, audio, and haptic channels. "Guide-LLM: An Embodied LLM Agent and Text-Based Topological Map for Robotic Guidance of People with Visual Impairments" (Song et al., 2024) separates a global, symbolic text-map planner from local camera-based localization, hazard reasoning, and low-level robot motion. "VeasyGuide: Personalized Visual Guidance for Low-vision Learners on Instructor Actions in Presentation Videos" (Sechayk et al., 29 Jul 2025) separates a low-level activity recognition pipeline from a high-level activity visualization module that turns detections into personalized highlights and zoom behavior (Sechayk et al., 29 Jul 2025).

Some systems define the two levels perceptually rather than computationally. "Bridging the Gap: Enhancing Gaze-Performance Link in Children with ASD through Dual-Level Visual Guidance in MR-DMT" (Liu et al., 4 Oct 2025) uses a perceptual level implemented by salience modulation and a transformational level implemented by kinetic metaphors such as animated fish that travel along limb trajectories. "Beyond Symbols: Motion Perception Cues Enhance Dual-Task Performance with Wearable Directional Guidance" (Zhang et al., 25 Jan 2026) prototypes a foveal task-focused channel together with a monocular, peripheral motion channel that carries directional guidance on a wearable display (Zhang et al., 25 Jan 2026). In these cases, the two levels are split across human perceptual pathways rather than across separate machine subsystems.

5. Empirical patterns and reported benefits

A recurring empirical claim is that decoupling levels improves spatial precision, robustness, interpretability, or latency. VP-VLA reports that experiments on the Robocasa-GR1-Tabletop benchmark and SimplerEnv simulation demonstrate that the method improves success rates by 5% and 8.3%, surpassing competitive baselines including QwenOFT and GR00T-N1.6 (Wang et al., 23 Mar 2026). On RoboCasa, ablations show that the full method reaches 53.8% average, compared with 49.4% without grounding loss, 49.5% with grounding at all frames, 47.3% when changing crosshair to point, and 50.8% for direct overlay onto the raw image (Wang et al., 23 Mar 2026).

ECVT reports 40.5% average mAP on ActivityNet v1.3 and 67.1% [email protected] on THUMOS14, outperforming leading baselines; ablations show that both GEP-only and TSEP-only improve over a baseline video transformer, while the full dual-level system performs best (Peng et al., 24 Aug 2025). DisCo-FLoc reports that it outperforms the state-of-the-art semantic-based method on two standard visual FLoc benchmarks, and ablations on Structured3D(full) show that adding both position-level and orientation-level negatives improves [email protected] m from 55.9 to 59.0 and R@1 m 30° from 63.2 to 66.0 (Meng et al., 5 Jan 2026).

In navigation, FSD-VLN reports up to 2X higher navigation success rates on unseen scenes than SOTA methods, while cutting single-action inference delay and total task runtime by over 50% (Zhu et al., 9 Jul 2026). DualVLN reports that the asynchronous design yields 0.03 s inference for 32 trajectories in parallel in System 1, and that it outperforms prior methods across VLN benchmarks while maintaining real-world long-horizon planning and dynamic adaptability (Wei et al., 9 Dec 2025). AerialVLA reports that on the TravelUAV benchmark it achieves state-of-the-art performance in seen environments and nearly three times the success rate of leading baselines in unseen scenarios, while an ablation shows that a 5-view design degrades unseen-map success relative to the streamlined dual-view setup (Xu et al., 15 Mar 2026).

In human-centered interfaces, effects are often reported as reduced search effort or improved dual-task performance. Guided Reality reports a user study with $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$5 and a reduction in error rate in a printer task from 17.14% to 10.42% (Zhao et al., 5 Aug 2025). VeasyGuide reports that low-vision learners improved in detecting instructor actions, with mean success rate increasing from 0.61 to 0.88 and a significant reduction in NASA-TLX dimensions such as mental demand and frustration (Sechayk et al., 29 Jul 2025). The wearable motion-cue study reports that, in a demanding dual-task scenario, the motion-based approach resulted in significantly more accurate interpretation of directional cues with $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$6 and showed a trend towards reduced errors on the concurrent primary task with $a_t = \pi_{\theta}(l, o_t, I_{vp}^t)$7 (Zhang et al., 25 Jan 2026). This suggests that dual-level guidance can improve performance either by redistributing computation across machine modules or by redistributing perceptual load across human channels.

6. Limitations, failure modes, and open directions

Across domains, the most common limitation is dependence on the correctness of the higher-level signal. VP-VLA states that performance depends on the correctness of System 2 reasoning and segmentation; if System 2 misidentifies objects or subtask progress, visual prompts will mislead the controller (Wang et al., 23 Mar 2026). ECVT notes dependence on LVLM quality and static offline guidance; if the LVLM misinterprets rare or novel actions, both global and local guidance can be misleading (Peng et al., 24 Aug 2025). DisCo-FLoc depends on accurate calibration and on the quality of candidate generation; repetitive layouts and orientation ambiguity remain challenging even after contrastive disambiguation (Meng et al., 5 Jan 2026).

Another limitation is restricted prompt expressiveness. VP-VLA notes that crosshair plus box are 2D and relatively coarse, and suggests richer affordance cues such as keypoints, heatmaps, or 3D prompts (Wang et al., 23 Mar 2026). DualVLN shows robustness to moderate pixel-goal errors but still fails on large or semantically incorrect goals, especially near obstacles (Wei et al., 9 Dec 2025). AerialVLA lacks global memory and can become conservative in highly repetitive environments, which suggests that purely reactive dual-view control may need an additional long-range memory or exploration mechanism (Xu et al., 15 Mar 2026).

In user-interface systems, personalization and ecological validity remain open issues. Guided Reality is currently iPad-based and open-loop, with no state-aware correction during rollout (Zhao et al., 5 Aug 2025). VeasyGuide uses a motion-based, content-agnostic detector; future systems could align highlights with semantic slide structures or support activity-based navigation (Sechayk et al., 29 Jul 2025). The ASD MR-DMT system uses static, rule-based guidance rather than real-time eye tracking and motion tracking; the paper explicitly points to adaptive and personalized guidance as future work (Liu et al., 4 Oct 2025).

Open directions stated across the literature converge on a few themes: more expressive visual prompts, tighter planner–controller co-training, dynamic prompting or online adaptation, richer graph or 3D scene representations, and stronger mechanisms for detecting unreliable high-level cues (Wang et al., 23 Mar 2026). A plausible implication is that the next generation of dual-level visual guidance systems will move from fixed, one-way guidance interfaces toward closed-loop systems in which the lower level not only executes but also informs, queries, or corrects the higher level.