An unstructured roadless environment navigation map construction method based on remote sensing

The capabilities of autonomous mobile robotic systems have been steadily improving due to recent advancements in computer science, engineering, and related disciplines such as cognitive science. In controlled environments, robots have achieved relatively high levels of autonomy. In more unstructured environments, however, the development of fully autonomous mobile robots remains challenging due to the complexity of understanding these environments. Many autonomous mobile robots use classical, learning-based or hybrid approaches for navigation. More recent learning-based methods may replace the complete navigation pipeline or selected stages of the classical approach. For effective deployment, autonomous robots must understand their external environments at a sophisticated level according to their intended applications. Therefore, in addition to robot perception, scene analysis and higher-level scene understanding (e.g., traversable/non-traversable, rough or smooth terrain, etc.) are required for autonomous robot navigation in unstructured outdoor environments. This paper provides a comprehensive review and critical analysis of these methods in the context of their applications to the problems of robot perception and scene understanding in unstructured environments and the related problems of localisation, environment mapping and path planning. State-of-the-art sensor fusion methods and multimodal scene understanding approaches are also discussed and evaluated within this context. The paper concludes with an in-depth discussion regarding the current state of the autonomous ground robot navigation challenge in unstructured outdoor environments and the most promising future research directions to overcome these challenges.

To address the local path planning problem for autonomous land vehicle in unstructured environments, a novel method combining navigation function and A* search is proposed. The computed navigation function in this paper embodies different constraints of vehicle, so in fact to be a near-feasible trajectory. Next the navigation function and sensed obstacles is used to construct the heuristic function, which is useful for the A* search of a local path from the initial state to the goal state in real time. Experiment results show that the proposed method can well solve the problem of local path planning for autonomous land vehicle in unstructured environments.

Automatic object searching is one of the essential skills for domestic robots to operate in unstructured human environments. It involves concatenation of several capabilities, including object identification, obstacle avoidance, path planning, and navigation. In this paper, we propose an automatic object searching framework for a mobile robot equipped with a single RGB-D camera. The obstacle avoidance is achieved by a behavior learning algorithm based on deep belief networks. The target object is recognized using scale-invariant feature transform descriptors and the relative position between the target and mobile robot is estimated from the RGB-D data. Subsequently, the mobile robot makes a path planning to the target location using an improved bug-based algorithm. The framework is tested in indoor environments and requires the robot to perform obstacle avoidance and automatically search and approach the target object. The results indicate that the system is collision free and reliable in performing searching tasks. This system’s functions make itself have the potential of being used for local navigation in unstructured environments.

<title>ABSTRACT</title> <p>Autonomous ground vehicles have the potential to reduce the risk to Soldiers in unfamiliar, unstructured environments. Unmanned operations in unstructured environments require the ability to guide the vehicles from their starting position to a target position. This paper proposes a framework to plan paths across such unstructured environments using a priori information about the environment as cost criteria into a multi-criteria, multi-agent path planner. The proposed multi-criteria, multi-agent path planner uses a penalty-based A* algorithm to plan multiple paths across the unstructured environment and uses entropy weighting for generating weights to calculate a multi-criteria cost with distance, risk, and soil trafficability. The paths generated by the proposed framework provide a better overall performance across the cost criteria and can be used as waypoints to navigate UGVs in off-road environments.</p> <p><bold>Citation:</bold> S. Khatiwada, P. Murray-Tuite, M.J. Schmid, “Multi-Criteria Multi-Agent Path Planning In Unstructured Off-Road Environments,” In <italic>Proceedings of the Ground Vehicle Systems Engineering and Technology Symposium</italic> (GVSETS), NDIA, Novi, MI, Aug. 15-17, 2023.</p>

The Map Navigation and Location Service is one of the most popular service types. To analyze the quality of the end-user experience, we identify the key indicators of quality of experience (QoE) for these services, and estimate the weights of these indicators. Based on the SERVPERF model, we collect the feedback data of college students and workers to build an index system of map navigation and location service. Using this index system, we can help users choose high-quality map navigation and location service, and to improve the end user experience for these services.KeywordsNavigation serviceQuality of experienceLocation service

This paper proposes a knowledge based genetic algorithm (GA) for path planning of a mobile robot in unstructured environments. The algorithm uses a unique problem representation method to represent 2-dimensional robot environments with complex obstacle layouts of arbitrary obstacle shapes. An effective evaluation method is specially developed for the proposed genetic algorithm. The evaluation method is able to accurately detect collisions between a robot path and an arbitrarily shaped obstacle, and assigns costs that are very effective for the proposed genetic algorithm. The proposed GA uses problem-specific genetic algorithms for robot path planning instead of the standard GAs. The proposed knowledge based genetic algorithm incorporates the domain knowledge of robot path planning into its specialized operators, where some also combine a local search technique. The effectiveness and efficiency of the proposed genetic algorithm is demonstrated by simulation studies. The irreplaceable role of the specialized genetic operators for solving robot path-planning problem is shown by a comparison study

A good performance for path planning is essential to carry out real-world missions. In this paper, the path planning consists of a global planner, that finds the optimal path, and in a local planner, recalculates the path to avoid obstacles. The main focus is to improve the performance of local planning techniques decreasing the complexity. There are two main ways to do it: bidirectional algorithms, improving time, and global planners, improving time and completeness. Thus, we propose a global planner algorithm that generates auxiliary nodes, backtracking by the goal node. We perform a comparison among A*, Bi A*, Artificial Potential Field (APF), Bi APF, Rapid Exploring Random Tree (RRT), and Bi RRT with and without the global planner through statistical metrics of time, path length, CPU, and memory. The results show the advantages of using bidirectional algorithms and the proposed global planner. The bidirectional algorithms decrease the time to return to the trajectory and sometimes assist in the algorithm's completeness. The proposed global planner reduced the planning time by 91.6% and improved the completeness of all algorithms in an unstructured indoor environment.

To address the local path planning problem for autonomous land vehicle in unstructured environments, this paper will incorporate the topological representation of the local map and a fast A∗ search. Topological connection of the local map will serve as a navigation function for the computing of heuristic value. To extract the topological representation, polar histogram is constructed iteratively. Outlets of temporal expanded state can be found by wave peak of polar histogram. To improve the efficiency of computing the heuristic value based on topological map, a lazy computing process is adopted. Experimental results show that the proposed method can well resolve the local path planning problem for autonomous land vehicle in unstructured environments, even if the localization of the vehicle may not be very accurate.

This paper presents a new path planning method for a mobile robot in an unstructured and dynamic environment. The method consists of two steps: first, a probabilistic roadmap (PRM) is constructed and stored as a graph whose nodes correspond to a collision-free world state for the robot; second, Q-learning-a method of reinforcement learning, is integrated with PRM to determine a proper path to reach the goal. In this manner, the robot is able to use past experience to improve its performance in avoiding not only static obstacles but also moving obstacles, without knowing the nature of the movements of the obstacles. The developed approach is applied to a simulated robot system. The results show that the hybrid PRM-Q path planner is able to converge to the right path successfully and rapidly.

Path planning systems using graph‐search algorithms such as A* usually operate in uniform plan‐view occupancy grids. However, the sensors used to construct these grids observe the environment in their own sample space based on sensor type and viewpoint. In this paper we present animage spacetechnique for path planning in unknown unstructured outdoor environments. Our method differs from previous techniques in that we perform path search directly in image space—the native sensor space of the imaging sensor. After an image space path has been found, it is used for navigation in the real world. By operating at the resolution of the image sensor, image space planning facilitates accurate robot vs. obstacle localization and enables a high degree of movement precision. Our image space planning techniques can potentially be used with many different kinds of sensor data, and we experimentally evaluate the use of stereo disparity and color information. We present an extension to the basic image space planning system called thecylindrical plannerthat simulates a 2π field of view with a cylindrically shaped occupancy grid. We believe that image space planning is well suited for use in the local subsystem of a hierarchical planner and implement ahybrid hierarchical plannerthat utilizes the cylindrical planner as a local planning subsystem and a two‐dimensional Cartesian planner as the global planning subsystem. All three systems are implemented and experimentally tested on a real robot. We evaluate the failure modes of image space planning and discuss how to avoid them. We find that image space enables precise real‐time, near‐field planning. © 2009 Wiley Periodicals, Inc.

Stone Aerospace is developing the SUNFISH® autonomous underwater vehicle which addresses many of the challenges of remote, autonomous exploration in unstructured environments. The AUV was designed to be a highly-capable platform for operating in a wide variety of complex 3D spaces, ranging from man-made (e.g. piers or harbors) to natural (e.g. reefs or caves). SUNFISH is a person-portable, 6-DOF hovering vehicle with built-in precision navigation and control, multibeam mapping, imaging, and CTD capabilities. Building on this base functionality7, we have developed high-level capabilities for performing simultaneous localization and mapping (SLAM), exploration, path planning, and precision docking. This paper describes these capabilities, and presents a demonstration in the unstructured labyrinthine 3D environment of Peacock Springs in northern Florida, USA. SUNFISH demonstrated the capability to autonomously explore this environment, creating a real-time map with which it navigated through the complex. SUNFISH penetrated 120 m into the caves and was able to autonomously find its way back to the deployment location. The capabilities developed for SUNFISH also have direct application to general inspection, monitoring, and survey tasks, as well as providing a basis for the types of autonomous vehicles that will explore where humans cannot go.

Accurate environment representation is one of the key challenges in autonomous ground vehicle navigation in unstructured environments. We propose a real-time optimization-based approach to terrain modeling and path planning in off-road and rough environments. Our method uses an irregular, hierarchical, graph-like environment model. A space-dividing tree is used to define a compact data structure capturing vertex positions and establishing connectivity. The same unique underlying data structure is used for both terrain modeling and path planning without memory reallocation. Local plans are generated by graph search algorithms and are continuously regenerated for on-the-fly obstacle avoidance inside the scope of the local terrain map. We show that implementing a hierarchical model over a regular space division reduces graph edge expansions by up to 84%. We illustrate the applicability of the method through experiments with an unmanned ground vehicle in both structured and unstructured environments.

Autonomous legged navigation in unstructured environments is still an open problem which requires the ability of an intelligent agent to detect and react to potential obstacles found in its area. These obstacles may range from vehicles, pedestrians, or immovable objects in a structured environment, like in highway or city navigation, to unpredictable static and dynamic obstacles in the case of navigating in an unstructured environment, such as a forest road. The latter scenario is usually more difficult to handle, due to the higher unpredictability. In this paper, we propose a vision dynamics approach to the path planning and navigation problem for a quadruped robot, which navigates in an unstructured environment, more specifically on a forest road. Our vision dynamics approach is based on a recurrent neural network that uses an RGB-D sensor as its source of data, constructing sequences of previous depth sensor observations and predicting future observations over a finite time span. We compare our approach with other state-of-the-art methods in obstacle-driven path planning algorithms and perform ablation studies to analyze the impact of architectural changes to our model components, demonstrating that our approach achieves superior performance in terms of successfully generating collision-free trajectories for the intelligent agent.

The environmental perception capability of ground unmanned vehicles in unstructured environments in the wild is the premise for their path planning in passable areas, and obstacle detection is an essential part of environmental perception technology. In a complex environment, water hazards in the driving section of ground unmanned vehicles will pose a significant threat to the driving safety of vehicles. For example, water ingress into the interior of an unmanned ground vehicle will cause serious faults such as short circuits and sensor failure. Therefore, the research on the detection technology of water hazards in the unstructured environment in the field is of great significance and value. In the unstructured environment in the wild, the texture features of water hazards are not apparent, the features are easily blocked by the shade of trees or other shadows, the proportion in the camera field of view is relatively small, and the recognition rate is low. This paper proposes a fast detection method for water hazards in unstructured environments in the wild based on image enhancement and improved anchor frame YOLOv3.

Robotic arm path planning in complex, unstructured environments often suffers from challenges such as excessive sampling randomness, low search efficiency, and redundant path nodes. To address these shortcomings, this paper introduces multi-mode dynamic sampling rapidly-exploring random tree (MMD-RRT), an enhanced path planning method based on the RRT algorithm. MMD-RRT dynamically adjusts its sampling strategy and step size in response to obstacles. Key improvements include: (1) a novel multi-mode dynamic sampling strategy that effectively reduces the randomness of sample points; (2) a dynamic window with variable step sizes designed to adapt to diverse environments and improve search efficiency; and (3) path optimization using a greedy strategy to eliminate redundant nodes. Experimental results show that, compared to traditional algorithms in unstructured environments, MMD-RRT reduces path length by an average of 24.89% and search time by an average of 75.90%. Validation using an XArm6 robotic arm confirms the algorithm’s ability to generate effective obstacle avoidance paths, ensuring the robot completes tasks safely and efficiently.