Autonomous vehicles (AV) are expected to improve, reshape, and revolutionize the future of ground transportation. It is anticipated that ordinary vehicles will one day be replaced with smart vehicles that are able to make decisions and perform driving tasks on their own. In order to achieve this objective, self-driving vehicles are equipped with sensors that are used to sense and perceive both their surroundings and the faraway environment, using further advances in communication technologies, such as 5G. In the meantime, local perception, as with human beings, will continue to be an effective means for controlling the vehicle at short range. In the other hand, extended perception allows for anticipation of distant events and produces smarter behavior to guide the vehicle to its destination while respecting a set of criteria (safety, energy management, traffic optimization, comfort). In spite of the remarkable advancements of sensor technologies in terms of their effectiveness and applicability for AV systems in recent years, sensors can still fail because of noise, ambient conditions, or manufacturing defects, among other factors; hence, it is not advisable to rely on a single sensor for any of the autonomous driving tasks. The practical solution is to incorporate multiple competitive and complementary sensors that work synergistically to overcome their individual shortcomings. This article provides a comprehensive review of the state-of-the-art methods utilized to improve the performance of AV systems in short-range or local vehicle environments. Specifically, it focuses on recent studies that use deep learning sensor fusion algorithms for perception, localization, and mapping. The article concludes by highlighting some of the current trends and possible future research directions.

Achieving precise state estimation is needed for the unmanned aerial vehicle to perform a successful flight with a high degree of stability. Nonetheless, obtaining accurate state estimation is considered challenging due to the inaccuracies associated with the measurements of the onboard commercial-off-the-shelf inertial measurement unit. The immense vibration of the vehicle's rotors makes these measurements suffer from issues like large drifts, biases, and immense unpredictable noise sequences. These issues cannot be significantly tackled using classical estimators, and an accurate sensor fusion technique needs to be developed. In this paper, a deep learning (DL) framework is developed to enhance the performance of the state estimator. A deep neural network (DNN) is trained using a deep-learning-based technique to identify the associated measurement noise models and filter them out. The dropout technique is adopted for training DNN to avoid overfitting and reduce the complexity of nets computations. Compared to the classical estimation results, the proposed DL technique demonstrates capabilities in identifying the measurement's noise characteristics. As an example, an enhancement in estimating the attitude states at near hover is proven using this approach. Furthermore, an actual hover flight was performed to validate the proposed estimation enhancement method.

This paper presents a custom designed in-pipe inspection robot that is developed for a pipe of diameter 0.203 m, commonly found in the oil and gas industry. Several pressure sensors are incorporated on board the robot, which are used for detecting leaks. The robot has a propeller arrangement, which not only drives the robot forwards, but also helps simulate a flow in an empty pipe, and thus aids detection of leaks. Further, the leak detection system is augmented by a neural network based verification framework, which improves the robustness of leak detection by allowing the operator to check their identification of a leak by passing it through a neural network based system. This paper presents the details of construction of the actual robot, and presents experimental data which shows successful neuralnetworks based detection of leaks in various scenarios.

This paper presents a low-cost based approach for solving the navigation problem of wheeled mobile robots to perform required tasks within indoor and outdoor environments. The presented solution is based on probabilistic approaches for multiple sensor fusion utilizing low-cost visual/inertial sensors. For the outdoor environment, the Extended Kalman Filter (EKF) is used to estimate the robot position and orientation, the system consists of wheel encoders, a reduced inertial sensor system (RISS), and a Global Positioning System (GPS). For the indoor environment, where GPS signals are blocked, another EKF algorithm is proposed using low cost depth sensor (Microsoft Kinect stream). EKF indoor localization is based on landmarks extracted from the depth measurements. A hybrid low-cost reduced navigation system (HLRNS) for indoor and outdoor environments is proposed and validated in both simulation and experimental environments. Additionally, an input-output state feedback linearization (I-O SFL) technique is used to control the robot to track the desired trajectory in such an environment. According to the conducted validation simulation and experimental testing, the proposed HLRNS provides an acceptable performance to be deployed for real-time applications.

In this paper, a modified algorithm of the Rapidly exploring Random Tree star (RRT*) is proposed. This method is called RRT*N. The main advantage of this method is its speed and robustness in finding the path to the target. An extension of this method to the 3D case is also presented and its capability of handling static obstacles and dynamic unknown moving obstacles in 2D and 3D environments, is shown. This improved method uses a probability distribution to generate new nodes. The nodes closest to the target have higher probability, which generates a tree centered on the line joining the robot to the target. It is shown that this method can be three times faster in finding the path to the target than the regular RRT* in the same environment. Simulation and experimental results are presented to show the robustness of the proposed RRT*N method. GRAPHICAL ABSTRACT