Amirmehdi Yazdani

This paper investigates the formation control of a disturbed air-ground multi-robot system consisting of unmanned aerial vehicles (UAVS) and unmanned ground vehicles (UGVs). An embedded collision-free formation control scheme, comprising two main components, is proposed. Firstly, a formation signal generator is developed based on distributed cooperative control approaches and artificial potential field-based methods. The generator, functioning as a virtual multi-agent system through a network of second-order integrator-chain dynamic nodes, is designed to “achieve” the formation control task while ensuring some specific constraints. Secondly, by embedding the corresponding node within each robot and taking its outputs as the tracking references, distinct constrained tracking controllers are designed for the UAVs and UGVs, respectively. Additionally, some finite-time disturbance observers are also employed to estimate and compensate for the lumped disturbances of UAVs and UGVs. With these two components, the proposed embedded formation control scheme completes the desired formation shape asymptotically for the air-ground UAVs-UGVs system under disturbances, while the collisions are successfully avoided between any two UAVs or two UGVs. Rigorous stability analysis and comparative simulations demonstrate the effectiveness of the proposed control scheme.

Accurate localization is a critical requirement for the successful deployment of mobile robots in indoor environments, particularly those characterized by symmetrical structures and features. Symmetrical indoor environments pose unique localization challenges due to the presence of repeated patterns and structures that can confound traditional localization methods. In such environments, accurately estimating the robot's pose relative to a global reference frame becomes increasingly challenging, leading to potential errors and inefficiencies in robot navigation and task execution. This paper presents an improved deep learning-based Monte Carlo localization (DMCL) framework for global localization of a mobile robot in symmetrical indoor environment using only 2D lidar. We first, converted 2D laser data to single channel 2D projected image and an occupancy grid. This 2D projected image is used to train the neural network to regress the 3DOF of robot. Finally, we integrated this trained neural network which estimate the robot pose in environment with MCL in the weight updating stage. The performance of both Monte Carlo Localization (MCL) and in DMCL methods in symmetrical indoor environment is investigated through extensive simulation studies. Verifying the effectiveness of the proposed method our network is able to obtain position accuracy of 0.15m and scene classification accuracy to 99% in simulation.

Multi-robot task allocation and coalition formation are critical challenges in robotics, essential for applications such as disaster response, search and rescue, environmental monitoring, exploration and mapping, surveillance and security, logistics, agriculture, military operations and healthcare. Therefore, it is essential to address these challenges and develop optimal solutions for implementing these concepts in real-world scenarios to effectively execute the previously mentioned applications. Hence, this paper presents a comprehensive survey and comparative analysis of different approaches for allocating tasks to multiple robots and forming coalitions to accomplish these tasks efficiently. The paper first provides a systematic categorization of the existing methods into four different groups namely behavior-based, market-based, optimization-based, and learning-based methods. Next, it analyzes the trade-off between different objectives, including minimizing task completion time, maximizing resource utilization, and balancing workload among robots. The paper also explores the impact of robot heterogeneity, task dependencies, and communication constraints on the performance of various algorithms. Furthermore, it discusses the challenges of dynamic task allocation and coalition formation in response to changes in the environment or robot failures.
Accordingly, the paper presents a comprehensive comparative study of the surveyed approaches, highlighting their substantial features including limitations and suitability for different application scenarios. As such, the paper identifies promising research directions, including the integration of machine learning techniques and the development of hybrid algorithms. Through this systematic analysis, the main aim is to provide researchers with a comprehensive understanding of the state-of-the-art in multi-robot task allocation and coalition formation, enabling them to select the most appropriate approach for their specific requirements.

The Australian and global agricultural industries are experiencing economic challenges. These challenges to farming are caused by a group of related factors including lack of labour supply, low profit margins, and high operational costs. Of all the factors impacting upon the operation and profitability of farms, it is human labour and chemical pesticides which are most effectual. One potential solution which has been proposed to reduce the impact of said factors is the use of robotic labour on farms. However, the available robotic crop disease classification models are lacking in accuracy and efficiency for this solution to be commercially viable. Therefore, this research aims to further understand what progress needs to be made in order to increase the accuracy and efficiency of crop disease classification models, specifically for disease species classification and disease severity classification purposes.
As part of this research, previous literature will be examined to determine the current scope of automatic intelligent crop disease classification capabilities. Subsequently, three State-of-the-Art (SotA) image classification models (AlexNet, MobileNetv3, and MLP-Mixer) will be trained and tested to determine the current standard of performance for crop disease species classification. Subsequently, this research will determine, based upon model performance of currently existing models, which model architecture produces favourable results for crop disease severity classification.

This paper presents the development of an efficient and smooth path planning algorithm tailored for autonomous systems operating in static environments. A modified bi-directional and goal-oriented rapidly-exploring random tree (RRT) algorithm is proposed, generating an initial rough global path quickly. To further enhance the quality of this path, we introduce a two-step optimization process, involving down-sampling to reduce redundant waypoints and up-sampling to improve path resolution. A cubic curve smoothing technique is then applied to ensure the path maintains continuity and remains collision-free, even in dense, clustered environments. The algorithm is validated by simulations in environments that approximate real-world conditions using MATLAB. This work primarily focuses on improving computational efficiency and path smoothness, addressing key challenges in robotic path planning.