GRU-Based Deep Learning Framework for Real-Time, Accurate, and Scalable UAV Trajectory Prediction

Accurate trajectory prediction is essential for ensuring flight safety, autonomous navigation, and efficient control of unmanned aerial vehicles (UAVs). However, existing neural network approaches often focus on a single platform, limiting their generalizability across UAVs with different dynamics. This study evaluates three neural architectures for spatiotemporal trajectory prediction using high-frequency flight data. Two UAV configurations, such as a fixed-wing (Cessna) and a multirotor (Phantom), were simulated in XPlane 11 under identical environmental conditions. Flight data of latitude, longitude, and altitude sampled at 80 Hz were preprocessed using a sliding-window segmentation method to preserve temporal dependencies. Backpropagation Neural Network (BPNN), Recurrent Neural Network (RNN), and Long Short-Term Memory (LSTM) models were trained using standardised preprocessing and tuned hyperparameters. Results show that recurrent models outperform the feedforward BPNN in capturing temporal patterns. For the Cessna dataset, the RNN achieved balanced accuracy with a mean-squared error (MSE) of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$5.5 \times 10^{-8}$</tex>, while the LSTM obtained the lowest MSE (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$2.7 \times 10^{-8}$</tex>). In the more dynamic Phantom dataset, the BPNN remained competitive (MSE <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\approx 1.8 \times 10^{-7}$</tex>), whereas RNN and LSTM maintained superior altitude prediction. These results highlight the suitability of lightweight recurrent architectures for real-time UAV trajectory prediction.

In this paper, the problem of using one active unmanned aerial vehicle (UAV) and four passive UAVs to localize a 3D target UAV in real time is investigated. In the considered model, each passive UAV receives reflection signals from the target UAV, which are initially transmitted by the active UAV. The received reflection signals allow each passive UAV to estimate the signal transmission distance which will be transmitted to a base station (BS) for the estimation of the position of the target UAV. Due to the movement of the target UAV, each active/passive UAV must optimize its trajectory to continuously localize the target UAV. Meanwhile, since the accuracy of the distance estimation depends on the signal-to-noise ratio of the transmission signals, the active UAV must optimize its transmit power. This problem is formulated as an optimization problem whose goal is to jointly optimize the transmit power of the active UAV and trajectories of both active and passive UAVs so as to maximize the target UAV positioning accuracy. To solve this problem, a Z function decomposition based reinforcement learning (ZD-RL) method is proposed. Compared to value function decomposition based RL (VD-RL), the proposed method can find the probability distribution of the sum of future rewards to accurately estimate the expected value of the sum of future rewards thus finding better transmit power of the active UAV and trajectories for both active and passive UAVs and improving target UAV positioning accuracy. Simulation results show that the proposed ZD-RL method can reduce the positioning errors by up to 39.4% and 64.6%, compared to VD-RL and independent deep RL methods, respectively.

The traditional leader-follower Unmanned Aerial Vehicle (UAV) formation cooperative positioning (CP) algorithm, based on relative ranging, requires at least four leader UAV positions to be known accurately, using relative distance with leader UAVs to achieve the unknown position follower UAV’s high-precision positioning. When the number of the known position leader UAVs is limited, the traditional CP algorithm is not applicable. Aiming at the minimum cooperative unit, which consists of a known position leader UAV and an unknown position follower UAV, this paper proposes a CP method based on the follower UAV’s moving vector. Considering the follower UAV can only acquire the single distance with the leader UAV at each distance-sampling period, it is difficult to determine the follower UAV’s spatial location. The follower UAV’s moving vector is used to construct position observation of the follower UAV’s inertial navigation system (INS). High-precision positioning is achieved by combining the follower UAV’s moving vector. In the process of CP, the leader UAV obtains a high-precision position by an INS/Global Positioning System (GPS) loosely integrated navigation system and transmits its position information to the follower UAV. Based on accurate modeling of the follower UAV’s INS, the position, velocity and heading observation equation of the follower UAV’s INS are constructed. The improved extended Kalman filtering is designed to estimate the state vector to improve the follower UAV’s positioning accuracy. In addition, considering that the datalink system based on radio signals may be interfered with by the external environment, it is difficult for the follower UAV to obtain relative distance information from the leader UAV in real time. In this paper, the availability of the relative distance information is judged by a two-state Markov chain. Finally, a real flight test is conducted to validate the performance of the proposed algorithm.

Occupant behavioral patterns, once extracted, could reveal cues about activities and space usage that could effectively get used for building systems to achieve energy savings. The ability to accurately predict the trajectories of occupants inside a room branched into different zones has many notable and compelling applications. For example - efficient space utilization and floor plans, intelligent building operations, crowd management, comfortable indoor environment, security, and evacuation or managing personnel. This paper proposes future occupant trajectory prediction using state-of-the-art time series prediction methods, i.e., Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models. These models are being implemented and compared to forecast occupant trajectories at a given time and location in a non-intrusive and reliable manner. The considered test-space for the collection of the dataset is a multi-utility area in an instrumented public building. The deployed 3D Stereo Vision Cameras capture the spatial location coordinates (x- and y- coordinates) from a bird's view angle without eliciting any other information that could reveal confidential data or uniquely identify a person. Our results showed that the GRU model forecasts were considerably more accurate than the LSTM model for the trajectory prediction. GRU prediction model achieved a Mean Squared Error (MSE) of 30.72 cm between actual and predicted location coordinates, and LSTM achieved an MSE of 47.13 cm, respectively, for multiple occupant trajectories within the monitored area. Another evaluation metric Mean Absolute Error (MAE) is used, and the GRU prediction model achieved an MAE of 3.14 cm, and the LSTM model achieved an MAE of 4.07 cm. The GRU model guarantees a high-fidelity occupant trajectory prediction for any given case with higher accuracy when compared to the baseline LSTM model.

In this paper, a mission planning system is presented that generates mission plans for a group of unmanned aerial vehicles (UAVs) to provide continuous surveillance over an urban area. Given the information of terrain and buildings in the target area, a two-stage approach is employed to solve the problem. In the first stage, a set of camera locations called the vantage set is generated that provides complete coverage of the target area. In the second stage, one or several UAVs are determined to collectively share the vantage set and their individual paths are generated to carry out the continuous surveillance duty. In both stages, evolutionary algorithms (genetic algorithm for vantage set generation and ant colony system for UAV/path planning) are used to search for the optimal solution. During the search, constraints such as the flying capabilities of UAVs and collision avoidance are imposed to guarantee the feasibility of the final result.

The rising number of small unmanned aerial vehicles (UAVs) expected in the next decade will enable a new series of commercial, service, and military operations in low altitude airspace as well as above densely populated areas. These operations may include on-demand delivery, medical transportation services, law enforcement operations, traffic surveillance and many more. Such unprecedented scenarios create the need for robust, efficient ways to monitor the UAV state in time to guarantee safety and mitigate contingencies throughout the operations. This work proposes a generalized monitoring and prediction methodology that utilizes realtime measurements of an autonomous UAV following a series of way-points. Two different methods, based on sinusoidal acceleration profiles and high-order splines, are utilized to generate the predicted path. The monitoring approach includes dynamic trajectory re-planning in the event of unexpected detour or hovering of the UAV during flight. It can be further extended to different vehicle types, to quantify uncertainty affecting the state variables, e.g., aerodynamic and other environmental effects, and can also be implemented to prognosticate safety-critical metrics which depend on the estimated flight path and required thrust. The proposed framework is implemented on a simplified, scalable UAV modeling and control system traversing 3D trajectories. Results presented include examples of real-time predictions of the UAV trajectories during flight and a critical analysis of the proposed scenarios under uncertainty constraints.

Cellular-enabled unmanned aerial vehicles (UAVs) require almost continuous cellular network connectivity to fulfill their missions successfully. However, the area (e.g., rural) they fly over may have partial coverage, making the path planning of such UAV missions a challenging task. Recently a tolerable outage duration is taken into account for such UAVs, and the trajectory optimization under this outage duration is studied. However, these existing studies consider only a single UAV and focus on optimization of each UAV's own path separately even in multi-UAV scenarios. In this paper, we study the trajectory optimization problem for cellular-enabled UAVs by taking into account the collaboration among UAVs. That is, for a given set of UAVs, each with a mission to fly from a starting point to an ending point, we aim to optimize the total mission completion time for all UAVs such that none of them has a connection outage more than a threshold. We let UAVs collaborate and provide connectivity as relays to each other to solve their outage problem and shorten their trajectories. We first model and solve this problem using nonlinear programming after discretization of the problem. Since it takes longer to solve the problem with such an approach, we then provide a graph-based approximate solution that runs fast. Numerical results show that the proposed approximate solution provides close to optimal results and performs better than state-of-the-art solutions that consider each UAV separately without collaboration among UAVs.

Unmanned aerial vehicle (UAV) flight trajectories in complex environments are often affected by multiple uncertainties, making accurate prediction challenging. To address this issue, this study proposes a probabilistic four-dimensional (4D) trajectory prediction model based on Brownian bridge motion. The UAV’s flight from mission start to endpoint is modeled as a Brownian bridge stochastic process with endpoint constraints, where the mean function sequence is constructed from path planning results and UAV performance parameters. To incorporate operational feasibility, the concept of the spatiotemporal reachable domain from time geography is introduced to dynamically constrain reachable positions, while a truncated Brownian bridge distribution is used to model probabilistic positions in three-dimensional space. A simulation platform in a realistic 3D geographical environment is developed to validate the model. Case studies show that the proposed method achieves dynamic probabilistic trajectory prediction under mission constraints with strong adaptability and practicality. The results provide theoretical support and technical reference for trajectory planning, conflict detection, and flight risk assessment in the pre-tactical phase.

Swarms of Unmanned Aerial Vehicles (UAV) have been foreseen by multiple organizations to serve an important role in future air-based warfare and civilian operations. UAVs are less expensive than their piloted counterparts, provide greater flexibilities and remove the need for on-board pilot support. Efficient control of swarms opens a set of new challenges, such as automatic UAV coordination, efficient swarm monitoring and dynamic mission planning. In this paper, we investigate the problem of dynamic mission planning for a UAV swarm. An agent-based control framework is proposed, which employs a control agent for task assignment and multiple UAV agents for local task scheduling. A prototype simulation framework is implemented as a proof-of-concept. Experimentation with the framework suggests the effectiveness of swarm control using several mission planning mechanisms.

Precise timing and precise location information are provided by Global Navigation Satellite Systems (GNSS) and play a crucial role in the positioning, navigation and data acquisition of most Unmanned Aerial Vehicles (UAV). GNSS functions include the following applications in UAVs: time-stamping and geo-referencing of collected data and images, synchronization of swarm flying and follow-me flights, determination of position and attitude in-flight, flight trajectory by following a pre-defined number of waypoints, mission planning, return home automatically without external control, avoidance of obstacles and geo-fencing. &#x0D; Some of these critical operations have implications for the safety of the UAV, the surrounding environment and health and safety of people, for example UAVs threatening to bring down aircrafts at airports, which are no-fly zones for UAVs. The appropriate GNSS based function to avoid this is geo-fencing. Another example is obstacle avoidance to prevent collisions and damages both for the UAV and the obstacle, e.g. anything from a window pane, tree, human being, to a power line.&#x0D; In order to ensure health and safety it is thus important to ensure correct function of the navigation and the timing, under a wide variety of circumstances, and in different signal environments. There can be signal disturbances, such as obscurations by buildings or reflected GNSS signals, called multipath. The performance of timing and navigation based on GPS/GNSS can be tested and verified in a controlled and repeatable way in the laboratory with different types of test equipment. We will give an introduction to a wide range of potential threats to GNSS Positioning, navigation and timing and an overview of different test methods. In addition, we are presenting a method for time synchronization of drones to enable safe swarm and follow flights in UAVs.

The unmanned aerial vehicle (UAV) has been applied in unmanned air combat because of its flexibility and practicality. The short-range air combat situation is rapidly changing, and the UAV has to make the autonomous maneuver decision as quickly as possible. In this paper, a type of short-range air combat maneuver decision method based on deep reinforcement learning is proposed. Firstly, the combat environment, including UAV motion model and the position and velocity relationships, is described. On this basic, the combat process is established. Secondly, some improved points based on proximal policy optimization (PPO) are proposed to enhance the maneuver decision-making ability. The gate recurrent unit (GRU) can help PPO make decisions with continuous timestep data. The actor network's input is the observation of UAV, however, the input of the critic network, named state, includes the blood values which cannot be observed directly. In addition, the action space with 15 basic actions and well-designed reward function are proposed to combine the air combat environment and PPO. In particular, the reward function is divided into dense reward, event reward and end-game reward to ensure the training feasibility. The training process is composed of three phases to shorten the training time. Finally, the designed maneuver decision method is verified through the ablation study and confrontment tests. The results show that the UAV with the proposed maneuver decision method can obtain an effective action policy to make a more flexible decision in air combat.

Neural networks (NNs) are now being extensively utilized in various artificial intelligence platforms specifically in the area of image classification and real-time object tracking. We propose a novel design to address the problem of real-time unmanned aerial vehicle (UAV) monitoring and detection using a Zynq UltraScale FPGA-based convolutional neural network (CNN). The biggest challenge while implementing real-time algorithms on FPGAs is the limited DSP hardware resources available on FPGA platforms. Our proposed design overcomes the challenge of autonomous real-time UAV detection and tracking using a Xilinx’s Zynq UltraScale XCZU9EG system on a chip (SoC) platform. Our proposed design explores and provides a solution for overcoming the challenge of limited floating-point resources while maintaining real-time performance. The solution consists of two modules: UAV tracking module and neural network–based UAV detection module. The tracking module uses our novel background-differencing algorithm, while the UAV detection is based on a modified CNN algorithm, designed to give the maximum field-programmable gate array (FPGA) performance. These two modules are designed to complement each other and enabled simultaneously to provide an enhanced real-time UAV detection for any given video input. The proposed system has been tested on real-life flying UAVs, achieving an accuracy of 82%, running at the full frame rate of the input camera for both tracking and neural network (NN) detection, achieving similar performance than an equivalent deep learning processor unit (DPU) with UltraScale FPGA-based HD video and tracking implementation but with lower resource utilization as shown by our results.

The use of unmanned aerial vehicles (UAVs) has been increasingly permeating diverse areas. Swarms of heterogeneous UAVs have the potential to accomplish richer tasks due to the combination of varied resources. However, controlling a swarm of heterogeneous UAVs poses interesting challenges, such as the optimal allocation of resources, the geographical distribution of the components of the swarm, the timely alternation of team members and the complete mission plan of the UAV swarm. In this chapter we explore the use of SwarmOS, a middleware for edge computing, focused on the cooperative communication of heterogeneous devices. The SwarmOS middleware acts as a communication mediator and facilitator, creating an organic network of devices. SwarmOS is currently used for decentralized and distributed Internet of Things (IoT) applications in a variety of platforms and protocols, and in the present work, we extend its use to the mission plans of UAV swarms. To evaluate the use of SwarmOS for UAV swarms, we selected some relevant applications from the rural realm, farming and livestock. Along with the SwarmOS middleware we propose a specification of mission plans for those applications, using a rigorous and verifiable model inspired in interaction protocol and soft institutions, which were previously used in hazard prevention of UAV. The specification of the use cases will be discussed and analyzed to pave the way for future implementations of the proposed mission plans.

In recent years, Unmanned Aerial Vehicle (UAV) technology has become more and more mature, and the demand for UAV s in all walks of life is also increasing. UAV formations or swarms have become an emerging organizational form. As one of the cores of multi-UAV cooperative missions, mission planning has attracted more and more researchers' attention. The multi-UAV dynamic mission assignment method based on the clustering algorithm proposed in this paper can realize the reasonable planning of the multi-UAV reconnaissance mission. First of all, this paper establishes a problem model of the reconnaissance mission. Then based on the trajectory prediction of targets, the improved k-means clustering algorithm is proposed to classify the mission targets. For the targets assigned to each UAV, we propose the dynamic Dijkstra algorithm to optimize the targets reconnaissance sequence. Finally, we designed a simulation experiment and verified the effectiveness of the algorithm proposed in this paper.

Using multiple unmanned aerial vehicles (UAVs) to perform some tasks cooperatively has received growing attention in recent years. Task assignment is a difficult problem in mission planning. Multiple tasks assignment problem for cooperating homogeneous UAVs is considered as a traditional combinatorial optimization problem. This paper addresses the problem of assigning multiple tasks to cooperative homogeneous UAVs, minimizing the total cost and balancing the cost of each UAV. We propose a centralized task assignment scheme which is based on minimum spanning tree. This scheme involves two phases. In the first phase, we use the Kruskal algorithm and the breadth first search algorithm to assign all tasks to UAVs and get a proper initial task assignment solution. The second phase involves the Pareto optimization improvement in the solution generated from the first phase. For a single UAV, we use the dynamic programming algorithm to calculate the total cost of completing all assigned tasks. The performance of the proposed scheme is compared to that of heuristic simulated annealing algorithm. The simulation results show that the proposed scheme can solve the homogeneous multi-UAV cooperative task assignment problem effectively.