Three-dimensional object detection utilizing LiDAR point cloud data is an indispensable part of autonomous driving perception systems. Point cloud-based 3D object detection has been a better replacement for higher accuracy than cameras during nighttime. However, most LiDAR-based 3D object methods work in a supervised manner, which means their state-of-the-art performance relies heavily on a large-scale and well-labeled dataset, while these annotated datasets could be expensive to obtain and only accessible in the limited scenario. Transfer learning is a promising approach to reduce the large-scale training datasets requirement, but existing transfer learning object detectors are primarily for 2D object detection rather than 3D. In this work, we utilize the 3D point cloud data more effectively by representing the birds-eye-view (BEV) scene and propose a transfer learning based point cloud semantic segmentation for 3D object detection. The proposed model minimizes the need for large-scale training datasets and consequently reduces the training time. First, a preprocessing stage filters the raw point cloud data to a BEV map within a specific field of view. Second, the transfer learning stage uses knowledge from the previously learned classification task (with more data for training) and generalizes the semantic segmentation-based 2D object detection task. Finally, 2D detection results from the BEV image have been back-projected into 3D in the postprocessing stage. We verify results on two datasets: the KITTI 3D object detection dataset and the Ouster LiDAR-64 dataset, thus demonstrating that the proposed method is highly competitive in terms of mean average precision (mAP up to $70\%$) while still running at more than 30 frames per second (FPS).

Autonomous navigation and collision avoidance missions represent a significant challenge for robotics systems as they generally operate in dynamic environments that require a high level of autonomy and flexible decision-making capabilities. This challenge becomes more applicable in micro aerial vehicles (MAVs) due to their limited size and computational power. This paper presents a novel approach for enabling a micro aerial vehicle system equipped with a laser range finder to autonomously navigate among obstacles and achieve a user-specified goal location in a GPS-denied environment, without the need for mapping or path planning. The proposed system uses an actor–critic-based reinforcement learning technique to train the aerial robot in a Gazebo simulator to perform a point-goal navigation task by directly mapping the noisy MAV’s state and laser scan measurements to continuous motion control. The obtained policy can perform collision-free flight in the real world while being trained entirely on a 3D simulator. Intensive simulations and real-time experiments were conducted and compared with a nonlinear model predictive control technique to show the generalization capabilities to new unseen environments, and robustness against localization noise. The obtained results demonstrate our system’s effectiveness in flying safely and reaching the desired points by planning smooth forward linear velocity and heading rates.

This paper introduces a novel design for achieving bearingless operation in a modular PMSM motor, specifically used for ultra-cleanroom applications in semiconductor wafer production. In order to achieve this operation, a stable and independent suspension of the rotor is crucial, which can be obtained through precise decoupling between the electromagnetic torque of the motor and the radial suspension force of the magnetic bearing assembly. To ensure this decoupling in the proposed machine, the stator of the motor part is isolated and distinct from the suspension winding poles named a suspension stator while the motoring rotor lies on the suspension rotor via an isolated material. Therefore, the independent placement of suspension force and torque assemblies ensures that the generated suspension force exhibits exceptional linearity concerning the rotor position and remains unaffected by variations in the torque current. This makes the suspension force more controllable for the levitation of the rotor assemblies in the air gap. To validate the effectiveness of this design, a mathematical model for the suspension force is established and analyzed through 3D finite element analysis (FEM). At the same time, the motor’s performance is investigated through 2D FEM. In addition, the control of the suspension assembly is conducted using Model Predictive Control (MPC) in MATLAB/Simulink. Finally, the influence of structural parameters on torque and suspension assemblies is analyzed in terms of decoupling characteristics using combined analysis via 3D FEM.

Quadrotor control in dynamic and unstructured environments requires robust and adaptive solutions to achieve agile and autonomous navigation. Nonlinear Model Predictive Control (NMPC) offers precise control through a nonlinear dynamical system model and online optimization over a short prediction horizon. However, its adaptability can be significantly enhanced through reinforcement learning. This study proposes a multimodal NMPC (MP-NMPC) policy, leveraging deep reinforcement learning to improve online optimization. Conditioned on the quadrotor's local observations, the trained reinforcement learning policy dynamically selects adaptation parameters and feedforward control commands for the low-level NMPC controller. The search for adaptation variables and feedforward control signals is formulated as a Markov Decision Process (MDP), enabling reinforcement learning to address this optimization challenge. A multimodal deep reinforcement learning policy that combines depth images with inertial data is trained in simulation and transferred to a real quadrotor in a zero-shot fashion, facilitating online hyperparameter adaptation in highly dynamic unstructured environments. The significance of this approach is demonstrated through its application to challenging tasks such as agile navigation and obstacle avoidance, where it enables autonomous flight through complex environments and outperforms standard reinforcement learning methods.

The increasing number of wildfires damages nature and human life, making the early detection of wildfires in complex outdoor environments critical. With the advancement of drones and remote sensing technology, infrared cameras have become essential for wildfire detection. However, as the demand for higher accuracy in detection algorithms grows, the detection model's size and computational costs increase, making it challenging to deploy high-precision detection algorithms on edge computing devices onboard drones for real-time fire detection. This paper introduces a novel infrared wildfire detection network named FCDNet to tackle this issue. It includes an Efficient Processing (EP) module based on the novel Partial Depthwise Convolution (PDWConv) and the lightweight feature-sharing decoupled detection head (Fast Head), achieving low-size and low-computation wildfire detection. An Adaptive Sample Attention (ASA) Loss is introduced to enhance the detection accuracy of wildfire cores in combination with Normalized Wasserstein Distance (NWD) Loss. The experiment shows that the model size of FCDNet is only 4.0MB, representing 63.5% of the baseline YOLOv8n network, with 63.3% of its parameters. It operates at just 5 Giga Floating Point Operations Per Second (GFLOPs), 38.3% lower, and achieves a 77.5% mAP (@50-95 IOU), a 1% increase, with a 460×460 input image size. Compared to the state-of-the-art YOLOv11n, FCDNet reduces parameters, computation, and model size by 26.9%, 20.6%, and 27.3%, respectively. The thermal dataset and training codes used in this study are made publicly available at: https://github.com/WangLF1996/FCDNet-Dataset-and-Algorithm.