Collision-Avoiding Flocking With Multiple Fixed-Wing UAVs in Obstacle-Cluttered Environments: A Task-Specific Curriculum-Based MADRL Approach.

Multiple unmanned aerial vehicles (UAVs) are able to efficiently accomplish a variety of tasks in complex scenarios. However, developing a collision-avoiding flocking policy for multiple fixed-wing UAVs is still challenging, especially in obstacle-cluttered environments. In this article, we propose a novel curriculum-based multiagent deep reinforcement learning (MADRL) approach called task-specific curriculum-based MADRL (TSCAL) to learn the decentralized flocking with obstacle avoidance policy for multiple fixed-wing UAVs. The core idea is to decompose the collision-avoiding flocking task into multiple subtasks and progressively increase the number of subtasks to be solved in a staged manner. Meanwhile, TSCAL iteratively alternates between the procedures of online learning and offline transfer. For online learning, we propose a hierarchical recurrent attention multiagent actor-critic (HRAMA) algorithm to learn the policies for the corresponding subtask(s) in each learning stage. For offline transfer, we develop two transfer mechanisms, i.e., model reload and buffer reuse, to transfer knowledge between two neighboring stages. A series of numerical simulations demonstrate the significant advantages of TSCAL in terms of policy optimality, sample efficiency, and learning stability. Finally, the high-fidelity hardware-in-the-loop (HITL) simulation is conducted to verify the adaptability of TSCAL. A video about the numerical and HITL simulations is available at https://youtu.be/R9yLJNYRIqY.

Actuator fault detection and isolation on multi-rotor UAV using extreme learning neuro-fuzzy systems.

Undetected partial actuator faults on multi-rotor UAVs can lead to system failures and uncontrolled crashes, necessitating the development of accurate and efficient fault detection and isolation (FDI) strategy. This paper proposes a hybrid FDI model for a quadrotor UAV that integrates an extreme learning neuro-fuzzy algorithm with a model-based extended Kalman filter (EKF). Three FDI models using Fuzzy-ELM, R-EL-ANFIS, and EL-ANFIS are compared based on training, validation performances, and sensitivity to weaker and shorter actuator faults. They are also tested online for linear and nonlinear incipient faults by measuring their isolation time delays and accuracies. The results show that the Fuzzy-ELM FDI model exhibits greater efficiency and sensitivity, while Fuzzy-ELM and R-EL-ANFIS FDI models demonstrate better performance than a conventional neuro-fuzzy algorithm, ANFIS.

Investigation of Using Sky Openness Ratio as Predictor for Navigation Performance in Urban-like Environment to Support PBN in UTM.

One of the causes of positioning inaccuracies in the Unmanned Aircraft System (UAS) is navigation error. In urban environment operations, multipaths could be the dominant contributor to navigation errors. This paper presents a study on how the operation environment affects the lateral (horizontal) navigation performance when a self-built UAS is going near different types of urban obstructions in real flight tests. Selected test sites are representative of urban environments, including open carparks, flight paths obstructed by buildings along one or both sides, changing sky access when flying towards corners formed by two buildings or dead ends, and buildings with reflective glass-clad surfaces. The data was analysed to obtain the horizontal position error between Global Positioning System (GPS) position and ground truth derived from Real Time Kinematics (RTK), with considerations for (1) horizontal position uncertainty estimate (EPH) reported by the GPS receiver, (2) no. of visible satellites, and (3) percentage of sky visible (or sky openness ratio, SOR) at various altitudes along the flight paths inside the aforementioned urban environments. The investigation showed that there is no direct correlation between the measured horizontal position error and the reported EPH; thus, the EPH could not be used for the purpose of monitoring navigation performance. The investigation further concluded that there is no universal correlation between the sky openness ratio (SOR) seen by the UAS and the resulting horizontal position error, and a more complex model would need to be considered to translate 3D urban models to expected horizontal navigation uncertainty for the UAS Traffic Management (UTM) airspace.

Discrete space-based route planning for rotary-wing UAV formation in urban environments.

The route planning problem for rotary-wing UAV formation is studied in discrete urban environments. First, a discrete-space consensus algorithm (DSCA) integrating asynchronous planning and grouping mechanism is developed to make the rotary-wing UAVs converge to the desired formation. Then the DSCA is combined with the improved rapidly-exploring random tree (IRRT) algorithm to enable the rotary-wing UAV formation to avoid the obstacles. Finally, the time information of waypoints is complemented by resolving the conflicts among multiple rotary-wing UAVs, thus reducing the rotary-wing UAVs' time gap of reaching the destinations. The proposed method extends the original consensus theory in the discrete space and for the obstacle avoidance issue. Also, it is valid in the route planning problem for rotary-wing UAV formation considering the real urban environments.

Hardware Development and Locomotion Control Strategy for an Over-Ground Gait Trainer: NaTUre-Gaits.

Therapist-assisted body weight supported (TABWS) gait rehabilitation was introduced two decades ago. The benefit of TABWS in functional recovery of walking in spinal cord injury and stroke patients has been demonstrated and reported. However, shortage of therapists, labor-intensiveness, and short duration of training are some limitations of this approach. To overcome these deficiencies, robotic-assisted gait rehabilitation systems have been suggested. These systems have gained attentions from researchers and clinical practitioner in recent years. To achieve the same objective, an over-ground gait rehabilitation system, NaTUre-gaits, was developed at the Nanyang Technological University. The design was based on a clinical approach to provide four main features, which are pelvic motion, body weight support, over-ground walking experience, and lower limb assistance. These features can be achieved by three main modules of NaTUre-gaits: 1) pelvic assistance mechanism, mobile platform, and robotic orthosis. Predefined gait patterns are required for a robotic assisted system to follow. In this paper, the gait pattern planning for NaTUre-gaits was accomplished by an individual-specific gait pattern prediction model. The model generates gait patterns that resemble natural gait patterns of the targeted subjects. The features of NaTUre-gaits have been demonstrated by walking trials with several subjects. The trials have been evaluated by therapists and doctors. The results show that 10-m walking trial with a reduction in manpower. The task-specific repetitive training approach and natural walking gait patterns were also successfully achieved.