No abstract

Along with the advancement of light-weight sensing and processing technologies, unmanned aerial vehicles (UAVs) have recently become popular platforms for intelligent traffic monitoring and control. UAV-mounted cameras can capture traffic-flow videos from various perspectives providing a comprehensive insight into road conditions. To analyze the traffic flow from remotely captured videos, a reliable and accurate vehicle detection-and-tracking approach is required. In this paper, we propose a deep-learning framework for vehicle detection and tracking from UAV videos for monitoring traffic flow in complex road structures. This approach is designed to be invariant to significant orientation and scale variations in the videos. The detection procedure is performed by fine-tuning a state-of-the-art object detector, You Only Look Once (YOLOv3), using several custom-labeled traffic datasets. Vehicle tracking is conducted following a tracking-by-detection paradigm, where deep appearance features are used for vehicle re-identification, and Kalman filtering is used for motion estimation. The proposed methodology is tested on a variety of real videos collected by UAVs under various conditions, e.g., in late afternoons with long vehicle shadows, in dawn with vehicles lights being on, over roundabouts and interchange roads where vehicle directions change considerably, and from various viewpoints where vehicles’ appearance undergo substantial perspective distortions. The proposed tracking-by-detection approach performs efficiently at 11 frames per second on color videos of 2720p resolution. Experiments demonstrated that high detection accuracy could be achieved with an average F1-score of 92.1%. Besides, the tracking technique performs accurately, with an average multiple-object tracking accuracy (MOTA) of 81.3%. The proposed approach also addressed the shortcomings of the state-of-the-art in multi-object tracking regarding frequent identity switching, resulting in a total of only one identity switch over every 305 tracked vehicles.

This paper proposes a high‐performance path following algorithm that combines Gaussian processes (GP) based learning and feedback linearization (FBL) with model predictive control (MPC) for ground mobile robots operating in off‐road terrains, referred to as GP‐FBLMPC. The algorithm uses a nominal kinematic model and learns unmodeled dynamics as GP models by using observation data collected during field experiments. Extensive outdoor experiments using a Clearpath Husky A200 mobile robot show that the proposed GP‐FBLMPC algorithm's performance is comparable to existing GP learning‐based nonlinear MPC (GP‐NMPC) methods with respect to the path following errors. The advantage of GP‐FBLMPC is that it is generalizable in reducing path following errors for different paths that are not included in the GP models training process, while GP‐NMPC methods only work well on exactly the same path on which GP models are trained. GP‐FBLMPC is also computationally more efficient than the GP‐NMPC because it does not conduct iterative optimization and requires fewer GP models to make predictions over the MPC prediction horizon loop at every time step. Field tests show the effectiveness and generalization of reducing path following errors of the GP‐FBLMPC algorithm. It requires little training data to perform GP modeling before it can be used to reduce path‐following errors for new, more complex paths on the same terrain (see video at https://youtu.be/tC09jJQ0OXM).