Search

In recent years, the use of multi-unmanned aerial vehicles (UAVs) systems in outdoor GPS enabled environments has increased significantly. Their use in GPS-denied environments, however, is still a complex task. Examples of such environments include search and rescue inside buildings or collapsed building scenarios, urban or natural canyons, in which the GPS signal is limited or blocked. In order to provide a possible solution for these applications, this paper presents a multi-UAV system that can be utilised for a team of UAVs to autonomously navigate, search and detect multiple targets in a cluttered and GPS-denied environment. The multi-UAV system uses a decentralised framework based on Decentralised Partially Observable Markov Decision Processes (Dec-POMDP) to formulate the decision-making process considering uncertainties in the environments and sensing. This system was designed for a mission that needs to cooperate efficiently to search and find multiple targets, without knowing possible locations, by sharing limited vital observations through WiFi connection. The system was tested in several cluttered and GPS-denied environments simulated in Gazebo and ROS interface. The environments were populated with different numbers and types of obstacles, several targets, a team of UAVs and boundaries. After comparing three different searching strategies, the results show that the system is capable of cooperatively searching and finding multiple targets in an effective approach while without having a collision in a complex environment. The multi-UAV system is robust to uncertainties in target exploration and localisation, enabling the use of multiple UAVs to enhance critical missions for challenging scenarios such as search and rescue or exploration and disaster management and assessment.

Unmanned Aerial Vehicles (UAVs) have been employed for multiple tasks in the last decade, including planetary exploration. A common task for UAVs is mapping, with maps most commonly represented in variable-resolution form, such as Octrees or their 2D-counterparts, Quadtrees. Convolutional Neural Networks (CNNs) have revolutionized sensor data processing, but are heavily influenced by the training data and the resolution of the images. In addition, Octree and Quadtree-based map representations typically insert values into the deepest level of the tree, regardless of the position of the UAV and the resulting image resolution. To alleviate both these issues, we propose a novel Quadtree implementation that considers the resolution of the observations when calculating insertion indices. The probabilistic mapping approach considers the tree structure when looking up probabilistic priors to insert generated observations. The system is implemented in ROS and tested in simulation and experiments with vertical, horizontal and mixed motions at different frequencies of observations. The map reproduction is evaluated with the Kullback-Leibler divergence and qualitatively. Increasing the frequency and approaching a target for closer inspection is shown to improve the quality of the map reproduction. Results also indicate that low-frequency, noisy observations can be overcome by increasing the frequency, opening up avenues for further research on mapping CNN observations. For a video demonstration visit: https://youtu.be/FdKu8BzjK0I

Planetary surfaces are a primary focus of space exploration. Some of the most challenging current efforts in planetary exploration relate to the search for life, or biosignatures, in these environments. Detecting water-related textures, and thus evidence for potentially habitable environments, has the potential to focus and accelerate the search for biosignatures on other planets. Desiccation cracks are sedimentary features providing evidence of sediment-water interaction. They are known from both Earth and Mars, and are likely to be found via aerial exploration approaches of ancient lakes, rivers, or shallow marine environments where biosignatures may be found. Current approaches using image processing to detect desiccation cracks focus on segmenting just the cracks and prove somewhat successful. However, the use of Unmanned Aerial Vehicles (UAVs) to detect and highlight areas with desiccation cracks for closer inspection over much larger surface areas has not yet been explored. This paper describes the development and deployment of a desiccation crack detection system using UAVs and AI. We describe data collection at varying heights above ground level and data-augmentation with a range of pixel-level and spatial transforms. Three state-of-the-art CNN segmentation networks are trained and evaluated using PyTorch. The networks are deployed on an edge-AI device integrated with a companion computer onboard a sub-2kg quadrotor UAV. Results indicate that the models can segment desiccation cracks on airborne-collected images at various locations at heights ranging from 5 to 20m. Deployment of the models for real-time inference onboard small UAVs shows potential for application in the field. This research shows the feasibility of a low-volume data training to UAV deployment pipeline while highlighting potential hurdles in the processing pipeline for future efforts. We present a system and architecture for onboard UAV detectors of sedimentary features, which can speed up t

Studies of past life forms in other planets are possible through the identification and localisation of desiccation cracks in ancient water bodies such as lakes, rivers and seas. Unmanned Aerial Vehicles (UAVs) are increasingly being used as a viable remote sensing solution as desiccation cracks are normally located in complex environments and are difficult to identify with the naked eye. However, as most UAVs have limited onboard decision-making capabilities for autonomous navigation in such environments, human operators have a strong reliance on their communication systems to operate them. This paper proposes a UAV-based system for autonomous onboard navigation, identification, and mapping of desiccation cracks for planetary exploration. The navigation problem is mathematically formulated as a Partially Observable Markov Decision Process (POMDP), where a motion strategy can be obtained by solving the POMDP using the Augmented Belief Tree (ABT) online solver. The UAV system is tested with Hardware in the Loop (HIL) simulations and real flight tests using two desiccation crack patterns distributed across an open area. Real-time segmentation from streamed camera frames of desiccation cracks is achieved through inference with a custom Convolution Neural Network (CNN) model that uses an architecture based on ResNet18, and an OpenCV AI Kit(OAK)-D camera. Real test results show that the system is capable of detecting and mapping the surveyed area with desiccation cracks effectively. The system design allows further adaptation for similar time-critical applications requiring increased levels of UAV autonomy in unstructured environments under uncertainty and partial observability such as humanitarian relief and surveillance.

The aim to colonize extra-terrestrial planets has been of great interest in recent years. For Mars, or other planets to sustain human life, it is essential that water is present and accessible. Mars may contain or provide signs of the possibility of near surface water, but this work will only focus on surface water for the purposes of simulation. In this paper, we present a method for autonomous navigation and detection of surface water bodies in the GPS denied environments of Mars via a fully autonomous UAV. A combination of existing state-of-the-art tools and techniques have been utilized to enable the development of this system. Additionally, we create a modular framework to simulate the mission using the AirSim simulation environment and the Robot Operating System (ROS). The simulation environment is leveraged by using the Unreal game engine running in Windows OS, interfacing with an open source software for simultaneous localization and mapping (SLAM), running on ROS and Linux OS. A simulated mission was successfully implemented and demonstrated using the framework. Results obtained indicate that the framework enables the navigation of a UAV in the simulated Mars environment and allows the UAV to detect surface water bodies. The developed simulation framework, along with the knowledge and techniques attained in this research, could accelerate the development, testing and deployment of missions for a real-world Mars UAV, for the detection of surface water bodies. Additionally, this research aims to support and build upon prior work to further aid the search for water on other planets, as well as assisting humans in becoming a multi-planet species.

The use of Unmanned Aerial Vehicles (UAVs) for Search And Rescue (SAR), powerlines, air quality and other applications is increasing. Their use has also been considered for planetary exploration (e.g. Mars, Titan). One exciting development in UAVs is a test planned by NASA of an unmanned helicopter in the atmosphere of Mars; this aims to establish a new dimension and direction for the planetary exploration field. Future missions will require advanced navigation tools supporting mission planning. The autonomy of UAVs systems will continue to grow for Earth applications supported by mathematical tools, models and formulations that help the UAV to deal with critical aspects of the mission. Planetary exploration is challenging and is influenced by different levels of uncertainty in UAV localization and the environment itself. Probabilistic navigation allows planning with uncertainty. This paper presents a high-level mission planning and navigation architecture for planetary exploration based on Partially Observable Markov Decision Process (POMDP). We focus on planetary exploration missions for biosignature detection. The paper presents a mission planning architecture and describes the results of a POMDP-based navigation and target finding module emulating biosignatures with ArUco markers in a Mars simulated environment.

The range of applications of unmanned aerial vehicles (UAVs) could be widened if a team of multiple UAVs are used. In this paper, we propose a framework of a team of UAVs with the aim of cooperatively finding a target in a real-world based environment with obstacles. Examples of such applications include search and rescue, remote sensing or infrastructure inspection, which can benefit from an efficient and cooperative multi-UAV system. The framework presented in this paper is modified and extended based on Partially Observable Markov Decision Processes (POMDP) to suit the decentralised multi-agent system while considering the necessary uncertainties of environments and localisations. In addition, the team can cooperate efficiently by sharing limited observation in the mission. We simulated the system in Gazebo simulator and tested the performances for an increased number of UAVs in a cluttered flying area. Results indicate that a POMDP formulation allows for uncertainty in observations and multiagent navigation and target finding can be implemented in a real-time application in real-world based scenarios.

In recent years the field of robotic navigation has increasingly harnessed semantic information in order to facilitate the planning and execution of robotic tasks. The use of semantic information focuses on employing representations more understandable by humans to accomplish tasks with robustness against environmental change, limiting memory requirements and improving scalability. Contemporary computer vision algorithms extracting semantic information have continuously improved their performance on benchmark datasets, however, most computations are expensive, limiting their use for robotic platforms constrained by size, weight and power such as unmanned aerial vehicles (UAVs), Recent advances have demonstrated the potential for navigation systems based on semantic information to be included into real-time operation of UAVs. This paper describes the development of a processing pipeline to incorporate the use of semantic information into a UAV navigation system. A navigation framework that uses the Robot Operating System (ROS) and semantic information is being developed, with simulations as a primary evaluation mechanism, preceeding deployment on hardware. The proposed system takes inputs from RGB images generated on-board the UAV and processes them in real-time to generate a semantic representation of its environment. The UAV executes subsequent actions autonomously by reasoning about the semantic content of the environment in order to accomplish a goal. Results from simulation indicated that the system is capable of extracting semantic information from camera images alone and infer plausible motion inputs for the flight controller to execute. The results also show that the system is capable of processing data in real-time and is able to enhance navigational capabilities to drive UAVs towards a higher level of autonomy.

Autonomous mission planning for unmanned aerial vehicles (UAVs) aims to leverage the capabilities of UAVs equipped with on-board sensors to accomplish a wide range of applications, including planetary exploration where greater science yields can be achieved at lower costs over shorter time periods. A significant body of research has already been performed with the aim of improving the autonomy of UAV missions, particularly in the areas of navigation and target identification. In this work, we review current approaches to drone navigation and exploration for planetary missions, with a focus on Mars and the main autonomy levels/techniques employed to achieve these levels. Recognising the importance of astrobiology in Mars exploration, we highlight progress in the area of autonomous biosignature detection capabilities trialed on Earth, and discuss the objectives and challenges in relation to future missions to Mars. Finally, we indicate currently available software tools and future work to improve autonomous mission planning capabilities.

Searching indoor environments in the presence of unknown obstacles with multiple UAV agents remains a challenge. This paper presents a framework for target-finding using a combination of traditional POMDP based planning and Deep Reinforcement Learning. The implementation breaks the problem into two separate stages of planning and control, with both stages modelled as a Partially Observable Markov Decision Process (POMDP). Global decentralised planning is provided using a modern online POMDP solver, while a modern Deep Reinforcement Learning algorithm is used to provide a policy for local control. Our results indicate that such a framework is capable of target-finding within a simulated indoor test environment in the presence of unknown obstacles, and once extended to real-world operation could enable UAVs to be applied in an increasing number of applications.

Response efforts in emergency applications such as border protection, humanitarian relief and disaster monitoring have improved with the use of Unmanned Aerial Vehicles (UAVs), which provide a flexibly deployed eye in the sky. These efforts have been further improved with advances in autonomous behaviours such as obstacle avoidance, take-off, landing, hovering and waypoint flight modes. However, most UAVs lack autonomous decision making for navigating in complex environments. This limitation creates a reliance on ground control stations to UAVs and, therefore, on their communication systems. The challenge is even more complex in indoor flight operations, where the strength of the Global Navigation Satellite System (GNSS) signals is absent or weak and compromises aircraft behaviour. This paper proposes a UAV framework for autonomous navigation to address uncertainty and partial observability from imperfect sensor readings in cluttered indoor scenarios. The framework design allocates the computing processes onboard the flight controller and companion computer of the UAV, allowing it to explore dangerous indoor areas without the supervision and physical presence of the human operator. The system is illustrated under a Search and Rescue (SAR) scenario to detect and locate victims inside a simulated office building. The navigation problem is modelled as a Partially Observable Markov Decision Process (POMDP) and solved in real time through the Augmented Belief Trees (ABT) algorithm. Data is collected using Hardware in the Loop (HIL) simulations and real flight tests. Experimental results show the robustness of the proposed framework to detect victims at various levels of location uncertainty. The proposed system ensures personal safety by letting the UAV to explore dangerous environments without the intervention of the human operator.

This paper presents a framework for UAV navigation in indoor environments using a deep reinforcement learning based approach. The implementation models the problem as two seperate problems, a Markov Decision Process (MDP), and a Partially Observable Markov Decision Processes (POMDP), separating the search problem into high-level planning and low-level action under uncertainty. We apply deep learning techniques to this layered problem to produce policies for the framework that allow a UAV to plan, act, and react. The approach is simulated and visualised using Gazebo and is evaluated using policies trained using deep-learning. Using recent deep-learning techniques as the basis of the framework, our results indicate that it is capable of providin smooth navigation for a simulated UAV agent exploring an indoor environment with uncertainty in its position. Once extended to real-world operation, this framework could enable UAVs to be applied in an increasing number of applications, from underground mining and oil refinery surveys and inspections, to search and rescue missions and biosecurity surveys.

Unmanned Aerial Vehicles (UAV) are increasingly used in a wide range of applications such as civil infrastructure inspection, agriculture, ecology and remote sensing. UAV autonomous operation relies on the use of GPS in order to localise and plan a mission, however there are places in which GPS positioning is limited or not available, such as in urban and natural canyons or below the canopy. This paper presents the development of a framework that enables a drone to navigate in an unknown, unstructured and GPS-denied environment. The aim of this research is to combine the use of localisation algorithms such as Simultaneous Localisation and Mapping (SLAM) with Partially Observable Markov Decision Processes (POMDP) algorithms into a framework in which the navigation and exploration tasks are modelled as sequential decision problems under uncertainty. The framework tested in simulation allows the UAV to navigate safely, avoiding collisions whilst guiding exploration in order to create an occupancy map of the UAV's surroundings. The proposed system guides the UAV through a series of actions in order to maximise the information gain about the unknown environment. The implementation of the proposed framework will enable the use of UAV for autonomous navigation and exploration in challenging environments where GPS positioning is not available or limited.

This paper describes field trials of Unmanned Aerial Vehicles (UAV) integrated with advanced digital hyperspectral and multispectral sensors to increase the efficiency of existing surveillance practices (human inspectors and insect traps) and application to detect a known endemic biosecurity pest (grape phylloxera) in Victorian vineyards. We evaluated airborne RGB, multi and hyperspectral imagery at two different vineyards with multiple grapevine varieties, in two separate time periods and under different levels of phylloxera infestation. The methods used to incorporate the sensors to the UAV, the flight operations and the processing workflow of the datasets from each imagery type are described. The ultimate aim of this study is to create an integrated methodology for collecting and processing multi and hyperspectral data with the purpose of remote sensing different variables in different applications such as, in this case, plant biosecurity. The development of a methodology for the collection and analysis of airborne multi and hyperspectral imagery would provide scientists with reliable data collection protocols and faster processing techniques to achieve different remote sensing objectives.

This paper presents a novel approach for mobile target tracking using a UAV flying in a cluttered and GPS-denied environment that occludes the target partially and momentarily. The problem is modelled as a sequential decision process under uncertainty using the Partially Observable Markov Decision Process framework (POMDP). The motion of the UAV is planned as sequences of high-level actions in order to optimise a multi-objective mission. A modular system executes the motion planning. The UAV and target dynamics models are included in the POMDP transition function as a black-box simulator. The target is detected using a front facing camera onboard the UAV. A model of the projection of the target and obstacles into the image plane of the camera is presented and used in the POMDP transition function. This model enables the POMDP solver to include the occlusions in the Field of View (FOV) of the camera caused by the obstacles in the scenario. Results in simulation indicate that the developed system is able to guide the UAV towards locations in which the target tracking duration is maximised whilst avoiding collisions with obstacles. This research has the potential to be used below the forest canopy, in fault or GPS-denied environments, underground mines and urban canyons.

Pest and disease detection is particularly challenging when managing crops that extend for hundreds or thousands of hectares. More efficient, targeted and accurate monitoring over such vast distances offers enormous benefits concerning monitoring and early detection of pests. Research carried out in CRC National Plant Biosecurity (CRCNPB) evaluated the potential of small Unmanned Aircraft Systems (sUAS) for deployment in plant biosecurity and provided government and industry with a series of recommendations. This project is a response to one of those recommendations: to investigate the application of sUAS to detect and monitor high priority in-field plant biosecurity threats. By combining modern digital photography with sUAS, the project team aimed to provide agricultural producers, resource managers, consultants, biosecurity personnel with the capacity to detect pests and diseases — including invasive species — before outbreaks occur. The technology is applicable across scales (plant-paddock-region), and can monitor across a range of host plants (e.g. wheat, vineyards, orchards) in diverse environments. Targeted pests and diseases including sugarcane aphid, cereal aphids, yellow stripe rust, myrtle rust and phylloxera were used to develop a generalized decision matrix to direct biosecurity surveillance programs to better predict the likelihood of pest presence and potential areas for surveillance. This matrix is transferrable to other pests and cropping systems. Training end-users in the use and on the capacity of sUAS occurred through several workshops, crop schools, and conferences in three countries, where seven remote pilots in command (RPIC) and over 20 biosecurity staff were formally trained.

Recent advances in remote sensed imagery and geospatial image processing using unmanned aerial vehicles (UAVs) have enabled the rapid and ongoing development of monitoring tools for crop management and the detection/surveillance of insect pests. This paper describes a (UAV) remote sensing-based methodology to increase the efficiency of existing surveillance practices (human inspectors and insect traps) for detecting pest infestations (e.g., grape phylloxera in vineyards).The methodology uses a UAV integrated with advanced digital hyperspectral, multispectral, and RGB sensors. We implemented the methodology for the development of a predictive model for phylloxera detection. In this method, we explore the combination of airborne RGB, multispectral,and hyperspectral imagery with ground-based data at two separate time periods and under different levels of phylloxera infestation. We describe the technology used—the sensors, the UAV, and the flight operations—the processing workflow of the datasets from each imagery type, and the methods for combining multiple airborne with ground-based datasets. Finally, we present relevant results of correlation between the different processed datasets. The objective of this research is to develop a novel methodology for collecting, processing, analysing and integrating multispectral, hyperspectral,ground and spatial data to remote sense different variables in different applications, such as, in this case, plant pest surveillance. The development of such methodology would provide researchers,agronomists, and UAV practitioners reliable data collection protocols and methods to achieve faster processing techniques and integrate multiple sources of data in diverse remote sensing applications.

Researchers have developed a novel remote sensing system using unmanned aerial vehicles, or drones, to maximise the accuracy and efficiency of monitoring efforts for detecting grape phylloxera pest infestations in vineyards.

Unmanned Aerial Vehicles (UAVs) are increasingly being used in numerous applications, such as remote sensing, environmental monitoring, ecology and search and rescue missions. Effective use of UAVs depends on the ability of the system to navigate in the mission scenario, especially if the UAV is required to navigate autonomously. There are particular scenarios in which UAV navigation faces challenges and risks. This creates the need for robust motion planning capable of overcoming different sources of uncertainty. One example is a UAV flying to search, track and follow a mobile ground target in GPS-denied space, such as below canopy or in between buildings, while avoiding obstacles. A UAV navigating under these conditions can be affected by uncertainties in its localisation and motion due to occlusion of GPS signals and the use of low cost sensors. Additionally, the presence of strong winds in the airspace can disturb the motion of the UAV. In this paper, we describe and flight test a novel formulation of a UAV mission for searching, tracking and following a mobile ground target. This mission is formulated as a Partially Observable Markov Decision Process (POMDP) and implemented in real flight using a modular framework. We modelled the UAV dynamic system, the uncertainties in motion and localisation of both the UAV and the target, and the wind disturbances. The framework computes a motion plan online for executing motion commands instead of flying to way-points to accomplish the mission. The system enables the UAV to plan its motion allowing it to execute information gathering actions to reduce uncertainty by detecting landmarks in the scenario, while making predictions of the mobile target trajectory and the wind speed based on observations. Results indicate that the system overcomes uncertainties in localisation of both the aircraft and the target, and avoids collisions into obstacles despite the presence of wind. This research has the potential of u

This research is a novel approach to enabling Unmanned Aerial Vehicle (UAV) navigation and target finding and tracking missions under uncertainty in cluttered and GPS-denied environments. A novel framework, implemented as a modular system, formulates the missions as online Partially Observable Markov Decision Processes (POMDP). The online POMDP computes a motion policy that balances multiple mission objectives optimally. The motion policy is updated while flying based onboard sensor observations. This research provides an enabling technology for UAV missions such as search and rescue, biodiversity assessment, underground mining and infrastructure inspection in challenging and natural environments.

In this paper we describe and flight test a novel system architecture for low cost muti-rotor unmanned aerial vehicles (UAVs) for searching, tracking and following a ground target. The UAV uses only on-board sensors for localisation within a GPS-denied space with obstacles. This mission is formulated as a Partially Observable Markov Decision Process (POMDP) and uses a modular framework that runs on the Robotic Operating System (ROS). This system computes a policy for executing actions instead of way-points to navigate and avoid obstacles. Results indicate that the system is robust to overcome uncertainties in localisation of both, the aircraft and the target and avoids collisions with the obstacles.

There are some scenarios in which Unmmaned Aerial Vehicle (UAV) navigation becomes a challenge due to the occlusion of GPS systems signal, the presence of obstacles and constraints in the space in which a UAV operates. An additional challenge is presented when a target whose location is unknown must be found within a confined space. In this paper we present a UAV navigation and target finding mission, modelled as a Partially Observable Markov Decision Process (POMDP) using a state-of-the-art online solver in a real scenario using a low cost commercial multi rotor UAV and a modular system architecture running under the Robotic Operative System (ROS). Using POMDP has several advantages to conventional approaches as they take into account uncertainties in sensor information. We present a framework for testing the mission with simulation tests and real flight tests in which we model the system dynamics and motion and perception uncertainties. The system uses a quad-copter aircraft with an board downwards looking camera without the need of GPS systems while avoiding obstacles within a confined area. Results indicate that the system has 100% success rate in simulation and 80% rate during flight test for finding targets located at different locations.

Unmanned Aerial Vehicles (UAV) can navigate with low risk in obstacle-free environments using ground control stations that plan a series of GPS waypoints as a path to follow. This GPS waypoint navigation does however become dangerous in environments where the GPS signal is faulty or is only present in some places and when the airspace is filled with obstacles. UAV navigation then becomes challenging because the UAV uses other sensors, which in turn generate uncertainty about its localisation and motion systems, especially if the UAV is a low cost platform. Additional uncertainty affects the mission when the UAV goal location is only partially known and can only be discovered by exploring and detecting a target. This navigation problem is established in this research as a Partially-Observable Markov Decision Process (POMDP), so as to produce a policy that maps a set of motion commands to belief states and observations. The policy is calculated and updated on-line while flying with a newly-developed system for UAV Uncertainty-Based Navigation (UBNAV), to navigate in cluttered and GPS-denied environments using observations and executing motion commands instead of waypoints. Experimental results in both simulation and real flight tests show that the UAV finds a path on-line to a region where it can explore and detect a target without colliding with obstacles. UBNAV provides a new method and an enabling technology for scientists to implement and test UAV navigation missions with uncertainty where targets must be detected using on-line POMDP in real flight scenarios.