The WikiRace game, where players navigate between Wikipedia articles using only hyperlinks, serves as a compelling benchmark for goal-directed search in complex information networks. This paper presents a systematic evaluation of navigation strategies for this task, comparing agents guided by graph-theoretic structure (betweenness centrality), semantic meaning (language model embeddings), and hybrid approaches. Through rigorous benchmarking on a large Wikipedia sub-graph, we demonstrate that a purely greedy agent guided by the semantic similarity of article titles is overwhelmingly effective. This strategy, when combined with a simple loop-avoidance mechanism, achieved a perfect success rate and navigated the network with an efficiency an order of magnitude better than structural or hybrid methods. Our findings highlight the critical limitations of purely structural heuristics for goal-directed search and underscore the transformative potential of large language models to act as powerful, zero-shot semantic navigators in complex information spaces.

While multi-vehicular collaborative driving demonstrates clear advantages over single-vehicle autonomy, traditional infrastructure-based V2X systems remain constrained by substantial deployment costs and the creation of "uncovered danger zones" in rural and suburban areas. We present AirV2X-Perception, a large-scale dataset that leverages Unmanned Aerial Vehicles (UAVs) as a flexible alternative or complement to fixed Road-Side Units (RSUs). Drones offer unique advantages over ground-based perception: complementary bird's-eye-views that reduce occlusions, dynamic positioning capabilities that enable hovering, patrolling, and escorting navigation rules, and significantly lower deployment costs compared to fixed infrastructure. Our dataset comprises 6.73 hours of drone-assisted driving scenarios across urban, suburban, and rural environments with varied weather and lighting conditions. The AirV2X-Perception dataset facilitates the development and standardized evaluation of Vehicle-to-Drone (V2D) algorithms, addressing a critical gap in the rapidly expanding field of aerial-assisted autonomous driving systems. The dataset and development kits are open-sourced at https://github.com/taco-group/AirV2X-Perception.

Unmanned Aerial Vehicles (UAVS) are limited by the onboard energy. Refinement of the navigation strategy directly affects both the flight velocity and the trajectory based on the adjustment of key parameters in the UAVS pipeline, thus reducing energy consumption. However, existing techniques tend to adopt static and conservative strategies in dynamic scenarios, leading to inefficient energy reduction. Dynamically adjusting the navigation strategy requires overcoming the challenges including the task pipeline interdependencies, the environmental-strategy correlations, and the selecting parameters. To solve the aforementioned problems, this paper proposes a method to dynamically adjust the navigation strategy of the UAVS by analyzing its dynamic characteristics and the temporal characteristics of the autonomous navigation pipeline, thereby reducing UAVS energy consumption in response to environmental changes. We compare our method with the baseline through hardware-in-the-loop (HIL) simulation and real-world experiments, showing our method 3.2X and 2.6X improvements in mission time, 2.4X and 1.6X improvements in energy, respectively.

-- We propose an opinion-driven navigation framework for multi-robot traversal through a narrow corridor . Our approach leverages a multi-agent decision-making model known as the Nonlinear Opinion Dynamics (NOD) to address the narrow corridor passage problem, formulated as a multi-robot navigation game. By integrating the NOD model with a multi-robot path planning algorithm, we demonstrate that the framework effectively reduces the likelihood of deadlocks during corridor traversal. T o ensure scalability with an increasing number of robots, we introduce a game reduction technique that enables efficient coordination in larger groups. Extensive simulation studies are conducted to validate the effectiveness of the proposed approach. In recent years, robots have become increasingly integrated into human environments, including residential areas, healthcare facilities, and public spaces. As robots interact frequently with both humans and other robots, the importance of social navigation has grown significantly. Social navigation focuses on optimizing a robot's behavior to enhance human comfort and improve the acceptability of robots in shared spaces. For instance, when multiple robots navigate through a narrow corridor, as depicted in Figure 1, they must dynamically adapt their movements in response to the actions of others, ensuring smooth and cooperative interactions within such constrained environments.

The increasing use of robots in human-centric public spaces such as shopping malls, sidewalks, and hospitals, requires understanding of how pedestrians respond to their presence. However, existing research lacks comprehensive datasets that capture the full range of pedestrian behaviors, e.g., including avoidance, neutrality, and attraction in the presence of robots. Such datasets can be used to effectively learn models capable of accurately predicting diverse responses of pedestrians to robot presence, which are crucial for advancing robot navigation strategies and optimizing pedestrian-aware motion planning. In this paper, we address these challenges by collecting a novel dataset of pedestrian motion in two outdoor locations under three distinct conditions, i.e., no robot presence, a stationary robot, and a moving robot. Thus, unlike existing datasets, ours explicitly encapsulates variations in pedestrian behavior across the different robot conditions. Using our dataset, we propose a novel Neural Social Robot Force Model (NSRFM), an extension of the traditional Social Force Model that integrates neural networks and robot-induced forces to better predict pedestrian behavior in the presence of robots. We validate the NSRFM by comparing its generated trajectories on different real-world datasets. Furthermore, we implemented it in simulation to enable the learning and benchmarking of robot navigation strategies based on their impact on pedestrian movement. Our results demonstrate the model's effectiveness in replicating real-world pedestrian reactions and its its utility in developing, evaluating, and benchmarking social robot navigation algorithms.

Autonomous Micro Air Vehicles (MAVs) are becoming essential in precision agriculture to enhance efficiency and reduce labor costs through targeted, real-time operations. However, existing unmanned systems often rely on GPS-based navigation, which is prone to inaccuracies in rural areas and limits flight paths to predefined routes, resulting in operational inefficiencies. To address these challenges, this paper presents ForaNav, an insect-inspired navigation strategy for autonomous navigation in plantations. The proposed method employs an enhanced Histogram of Oriented Gradient (HOG)-based tree detection approach, integrating hue-saturation histograms and global HOG feature variance with hierarchical HOG extraction to distinguish oil palm trees from visually similar objects. Inspired by insect foraging behavior, the MAV dynamically adjusts its path based on detected trees and employs a recovery mechanism to stay on course if a target is temporarily lost. We demonstrate that our detection method generalizes well to different tree types while maintaining lower CPU usage, lower temperature, and higher FPS than lightweight deep learning models, making it well-suited for real-time applications. Flight test results across diverse real-world scenarios show that the MAV successfully detects and approaches all trees without prior tree location, validating its effectiveness for agricultural automation.

The advancement in autonomous vehicles has empowered navigation and exploration in unknown environments. Geomagnetic navigation for autonomous vehicles has drawn increasing attention with its independence from GPS or inertial navigation devices. While geomagnetic navigation approaches have been extensively investigated, the generalizability of learned geomagnetic navigation strategies remains unexplored. The performance of a learned strategy can degrade outside of its source domain where the strategy is learned, due to a lack of knowledge about the geomagnetic characteristics in newly entered areas. This paper explores the generalization of learned geomagnetic navigation strategies via deep reinforcement learning (DRL). Particularly, we employ DRL agents to learn multiple teacher models from distributed domains that represent dispersed navigation strategies, and amalgamate the teacher models for generalizability across navigation areas. We design a reward shaping mechanism in training teacher models where we integrate both potential-based and intrinsic-motivated rewards. The designed reward shaping can enhance the exploration efficiency of the DRL agent and improve the representation of the teacher models. Upon the gained teacher models, we employ multi-teacher policy distillation to merge the policies learned by individual teachers, leading to a navigation strategy with generalizability across navigation domains. We conduct numerical simulations, and the results demonstrate an effective transfer of the learned DRL model from a source domain to new navigation areas. Compared to existing evolutionary-based geomagnetic navigation methods, our approach provides superior performance in terms of navigation length, duration, heading deviation, and success rate in cross-domain navigation. Geomagnetic navigation leverages the ubiquitous earth magnetic field signals for the navigation [1], [2], without independence on dedicated devices along the navigation route [3]-[5]. Geomagnetic navigation thus can secure the navigation mission, e.g., in remote areas or underwater environments where there GPS or pre-deployed navigation devices is unavailable [6].

In daily domestic settings, frequently used objects like cups often have unfixed positions and multiple instances within the same category, and their carriers frequently change as well. As a result, it becomes challenging for a robot to efficiently navigate to a specific instance. To tackle this challenge, the robot must capture and update scene changes and plans continuously. However, current object navigation approaches primarily focus on the semantic level and lack the ability to dynamically update scene representation. In contrast, this paper captures the relationships between frequently used objects and their static carriers. It constructs an open-vocabulary Carrier-Relationship Scene Graph (CRSG) and updates the carrying status during robot navigation to reflect the dynamic changes of the scene. Based on the CRSG, we further propose an instance navigation strategy that models the navigation process as a Markov Decision Process. At each step, decisions are informed by the Large Language Model's commonsense knowledge and visual-language feature similarity. We designed a series of long-sequence navigation tasks for frequently used everyday items in the Habitat simulator. The results demonstrate that by updating the CRSG, the robot can efficiently navigate to moved targets. Additionally, we deployed our algorithm on a real robot and validated its practical effectiveness. The project page can be found here: https://OpenIN-nav.github.io.

In the futuristic agricultural fields compatible with Agriculture 4.0, robots are envisaged to navigate through crops to perform functions like pesticide spraying and fruit harvesting, which are complex tasks due to factors such as non-geometric internal obstacles, space constraints, and outdoor conditions. In this paper, we attempt to employ Deep Reinforcement Learning (DRL) to solve the problem of 4WIS4WID mobile robot navigation in a structured, automated agricultural field. This paper consists of three sections: parameterization of four-wheel steering configurations, crop row tracking using DRL, and autonomous navigation of 4WIS4WID mobile robot using DRL through multiple crop rows. We show how to parametrize various configurations of four-wheel steering to two variables. This includes symmetric four-wheel steering, zero-turn, and an additional steering configuration that allows the 4WIS4WID mobile robot to move laterally. Using DRL, we also followed an irregularly shaped crop row with symmetric four-wheel steering. In the multiple crop row simulation environment, with the help of waypoints, we effectively performed point-to-point navigation. Finally, a comparative analysis of various DRL algorithms that use continuous actions was carried out.

Autonomous navigation guided by natural language instructions is essential for improving human-robot interaction and enabling complex operations in dynamic environments. While large language models (LLMs) are not inherently designed for planning, they can significantly enhance planning efficiency by providing guidance and informing constraints to ensure safety. This paper introduces a planning framework that integrates LLMs with 2D occupancy grid maps and natural language commands to improve spatial reasoning and task execution in resource-limited settings. By decomposing high-level commands and real-time environmental data, the system generates structured navigation plans for pick-and-place tasks, including obstacle avoidance, goal prioritization, and adaptive behaviors. The framework dynamically recalculates paths to address environmental changes and aligns with implicit social norms for seamless human-robot interaction. Our results demonstrates the potential of LLMs to design context-aware system to enhance navigation efficiency and safety in industrial and dynamic environments.