Autonomous Vehicle Tracking Research Articles

Deep-Learning-Based Floor Path Model for Route Tracking of Autonomous Vehicles

Real-time route tracking is an important research topic for autonomous vehicles used in industrial facilities. Traditional methods such as copper line tracking on the ground, wireless guidance systems, and laser systems are still used in route tracking. In this study, a deep-learning-based floor path model for route tracking of autonomous vehicles is proposed. A deep-learning floor path model and algorithm have been developed for highly accurate route tracking, which avoids collisions of vehicles and follows the shortest route to reach the destination. The floor path model consists of markers. Routes in the floor path model are created by using these markers. The floor path model is transmitted to autonomous vehicles as a vector by a central server. The server dispatches the target marker address to the vehicle to move. The vehicle calculates all possible routes to this address and chooses the shortest one. Marker images on the selected route are processed using image processing and classified with a pre-trained deep-CNN model. If the classified image and the image on the selected route are the same, the vehicle proceeds toward its destination. While the vehicle moves on the route, it sends the last classified marker to the server. Other autonomous vehicles use this marker to determine the location of this vehicle. Other vehicles on the route wait to avoid a collision. As a result of the experimental studies we have carried out, the route tracking of the vehicles has been successfully achieved.

Bias-Learning-Based Model Predictive Controller Design for Reliable Path Tracking of Autonomous Vehicles Under Model and Environmental Uncertainty

Abstract Path tracking error control is an essential functionality in the development of autonomous vehicles to follow a planned trajectory. Significant path tracking errors could lead to a collision or even out of the control of the vehicle. Model-based control strategies have been developed to minimize the vehicle’s path tracking errors. However, the vehicle model may not truly represent the actual vehicle dynamics. Furthermore, the parameters employed in the vehicle dynamic model may not represent the actual operating conditions of the vehicle under environmental uncertainty. This paper proposes a real-time bias-learning method coupling with the model predictive control (MPC) to improve the fidelity of a baseline vehicle model with the aid of a few experiments (or virtual experiments) so that the path tracking error can be reduced in real-time operation. Gaussian process (GP) regression and recurrent neural network (RNN) are employed for bias-learning and their effectiveness are compared under different scenarios. GP regression learns non-linearity of the model bias through its nonlinear kernel function, whereas the RNN model formulates the bias as a linear combination of hidden nodes which capture the non-linearity of the model bias with a recurrent form. Results reveal that RNN is more effective for real-time learning of the nonlinear model bias than the classical GP regression and the proposed bias-learning model is able to improve the fidelity of a baseline vehicle dynamic model. Consequently, path tracking performance can be greatly improved under environmental uncertainty using the bias-learning-based MPC.

Model predictive control with fuzzy logic switching for path tracking of autonomous vehicles

This paper introduces an integrated path tracking control strategy for autonomous vehicles. The proposed control strategy is based on a multi-input multi-output linear model predictive control (LMPC) with a fuzzy logic switching system. The designed MPC is based on Laguerre networks. The main target of the designed MPC is to produce the optimal control signals of the steering angle and the angular velocity while considering the physical constraints of the control signals and the measurements noise. Since the vehicle model is highly nonlinear and is operated over a wide range of operating points, different linearized models are obtained. The controller parameters for each linear model are designed and tuned. The gab metric analysis is used to select a number of these models to simplify the design of the proposed controller. Then, these models are combined using a fuzzy logic controller to switch between them. To test the proposed controller performance, different paths are generated using path planning algorithms. These paths simulate different vehicle maneuvers scenarios. The simulation results show that the designed tracking controller has a tracking performance on different designed paths better than that of a Linear quadratic gaussian (LQG) tracking controller, discussed in this paper.

MPC Based Trajectory Tracking for An Automonous Deep-Sea Tracked Mining Vehicle

Model predictive control (MPC) has been successfully used in trajectory tracking for autonomous vehicles based on certain kinematic model under low external disturbance conditions, but when there are model uncertainties and external disturbances, autonomous vehicles will fail to follow the pre-set trajectory. This paper studies trajectory tracking control based on MPC for an autonomous deep-sea tracked mining vehicle in polymetallic nodule mines with model uncertainty and external disturbances. A MPC algorithm is designed for trajectory tracking. To address model uncertainties caused by vehicle body subsidence and track slippage, a drive wheel speed correction controller is designed by experimental data fitting, and Kalman filtering (KF) and adaptive Kalman filtering (AKF) are introduced to improve tracking performance by rejecting external disturbances especially during curve tracking. To handle dead zones and obstacles during actual operation, an obstacle avoidance strategy is proposed that uses the tri-circular arc obstacle avoidance trajectory with an equal curvature for path re-planning. Finally, Simulink&Recurdyn co-simulations validate the performance of the proposed MPC controller through a comparison with nonlinear MPC(NMPC).

Review article: State-of-the-art trajectory tracking of autonomous vehicles

Abstract. Air pollution, energy consumption, and human safety issues have aroused people's concern around the world. This phenomenon could be significantly alleviated with the development of automatic driving techniques, artificial intelligence, and computer science. Autonomous vehicles can be generally modularized as environment perception, path planning, and trajectory tracking. Trajectory tracking is a fundamental part of autonomous vehicles which controls the autonomous vehicles effectively and stably to track the reference trajectory that is predetermined by the path planning module. In this paper, a review of the state-of-the-art trajectory tracking of autonomous vehicles is presented. Both the trajectory tracking methods and the most commonly used trajectory tracking controllers of autonomous vehicles, besides state-of-art research studies of these controllers, are described.

Path tracking based on model predictive control with variable predictive horizon

Model predictive control is one of the main methods used in path tracking for autonomous vehicles. To improve the path tracking performance of the vehicle, a path tracking method based on model predictive control with variable predictive horizon is proposed in this paper. Based on the designed model predictive controller for path tracking, the response analysis of path tracking control system under the different predictive horizons is carried out to clarify the influence of predictive horizon on path tracking accuracy, driving comfort and real-time of the control algorithm. Then, taking the lateral offset, the steering frequency and the real-time of the control algorithm as comprehensive performance indexes, the particle swarm optimization algorithm is designed to realize the adaptive optimization for the predictive horizon. The effectiveness of the proposed method is evaluated via numerical simulation based on Simulink/CarSim and hardware-in-the-loop experiment on an autonomous driving simulator. The obtained results show that the optimized predictive horizon can adapt to the different driving environment, and the proposed path tracking method has good comprehensive performance in terms of path tracking accuracy of the vehicle, driving comfort and real-time.

A novel Siamese Attention Network for visual object tracking of autonomous vehicles

In the environment perception stage of autonomous driving, vehicles need to track its surrounding objects quickly and accurately to avoid dangerous behaviors. Therefore, visual object tracking has important practical application value in autonomous driving system. However, the performance of most hierarchical convolutional feature trackers are limited by ignoring the complex environment of autonomous driving. In this paper, a novel Siamese Attention Network to explore the rich spatial and channel information of objects was proposed. Because of the lack of important information between the channel and the spatial position, the tracking performance is reduced by the challenges of illumination change and deformation. The spatial attention block and channel attention block focus on the importance of different spatial positions and channels, respectively. The effective fusion of the two makes our tracker achieve the state-of-the art performance of 0.300 in the EAO criterion of 2017, which exceeds the baseline by 5.7%.

Semantic Point Cloud-Based Adaptive Multiple Object Detection and Tracking for Autonomous Vehicles

LiDAR-based Multiple Object Detection and Tracking (MODT) is one of the essential tasks in autonomous driving. Since MODT is directly related to the safety of an autonomous vehicle, it is critical to provide reliable information about the surrounding objects. For that reason, we propose a semantic point cloud-based adaptive MODT system for autonomous driving. Semantic point clouds emerge with advances in deep learning-based Point Cloud Semantic Segmentation (PCSS), which assigns semantic information to each point in the point cloud of LiDAR. This semantic information provides several advantages to the MODT system. First, any point corresponding to any static object can be filtered. Because the class information assigned to each point can be directly utilized, filtering is possible without any modeling. Second, the class information of an object can be inferred without any special classification process because the class information is provided from the semantic point cloud. Finally, the clustering and tracking module can consider unique dimensional and dynamic characteristics based on class information. We utilize the Carla simulator and KITTI dataset to verify our method by comparing several existing algorithms. In conclusion, the performance of the proposed algorithm is improved by about 176% on average compared to the existing algorithm.

Computer vision based obstacle detection and target tracking for autonomous vehicles

Obstacle detection and target tracking are two major issues for intelligent autonomous vehicles. This paper proposes a new scheme to achieve target tracking and real-time obstacle detection of obstacles based on computer vision. ResNet-18 deep learning neural network is utilized for obstacle detection and Yolo-v3 deep learning neural network is employed for real-time target tracking. These two trained models can be deployed on an autonomous vehicle equipped with an NVIDIA Jetson Nano motherboard. The autonomous vehicle moves to avoid obstacles and follow tracked targets by camera. Adjusting the steering and movement of the autonomous vehicle according to the PID algorithm during the movement, therefore, will help the proposed vehicle achieve stable and precise tracking.

Autonomous vehicle tracking algorithm for high curvature path in full speed range

Autonomous vehicle tracking algorithm for high curvature path in full speed range

Kalman-filter-based sensor fusion applied to road-objects detection and tracking for autonomous vehicles

In this article, a real-time road-Object Detection and Tracking (LR_ODT) method for autonomous driving is proposed. This method is based on the fusion of lidar and radar measurement data, where they are installed on the ego car, and a customized Unscented Kalman Filter is employed for their data fusion. The merits of both devices are combined using the proposed fusion approach to precisely provide both pose and velocity information for objects moving in roads around the ego car. Unlike other detection and tracking approaches, the balanced treatment of both pose estimation accuracy and its real-time performance is the main contribution in this work. The proposed technique is implemented using the high-performance language C++ and utilizes highly optimized math and optimization libraries for best real-time performance. Simulation studies have been carried out to evaluate the performance of the LR_ODT for tracking bicycles, cars, and pedestrians. Moreover, the performance of the Unscented Kalman Filter fusion is compared to that of the Extended Kalman Filter fusion showing its superiority. The Unscented Kalman Filter has outperformed the Extended Kalman Filter on all test cases and all the state variable levels (−24% average Root Mean Squared Error). The employed fusion technique shows how outstanding is the improvement in tracking performance compared to the use of a single device (−29% Root Mean Squared Error with lidar and −38% Root Mean Squared Error with radar).

Implementation of MPC-Based Path Tracking for Autonomous Vehicles Considering Three Vehicle Dynamics Models with Different Fidelities

Model predictive control (MPC) algorithm is established based on a mathematical model of a plant to forecast the system behavior and optimize the current control move, thus producing the best future performance. Hence, models are core to every form of MPC. An MPC-based controller for path tracking is implemented using a lower-fidelity vehicle model to control a higher-fidelity vehicle model. The vehicle models include a bicycle model, an 8-DOF model, and a 14-DOF model, and the reference paths include a straight line and a circle. In the MPC-based controller, the model is linearized and discretized for state prediction; the tracking is conducted to obtain the heading angle and the lateral position of the vehicle center of mass in inertial coordinates. The output responses are discussed and compared between the developed vehicle dynamics models and the CarSim model with three different steering input signals. The simulation results exhibit good path-tracking performance of the proposed MPC-based controller for different complexity vehicle models, and the controller with high-fidelity model performs better than that with low-fidelity model during trajectory tracking.

Real-Time Adaptive Object Detection and Tracking for Autonomous Vehicles

The recent development of deep neural networks attempts to go deeper through the layered architecture to solve complex problems. As expected, the deepening impacts the processing times for inferences in tasks like object classification and localization. This demands solutions for balancing computational resource requirements and extend the versatility of a given model to suit a variety of applications. Often the solutions rely on architectural modifications to ease training times while achieving more accurate results. In this work, we propose a multistage algorithm with an initial stage adopting the YOLOv3 network for object detection and HART approach for object tracking. This is followed by an adaptive post-localization stage shift system taking into consideration the processing times of stage inferences, which are the number of located objects in image sequences. The goal is to change the localization strategy to achieve optimal processing time performance. For evaluation, we present results in terms of precision and processing times in varying traffic conditions. The results demonstrate the effectiveness of the proposed adaptive multistage model in comparison to other real-time state-of-the-art detection strategies, as it achieves a frame rate gain of 49% over YOLOv3 while maintaining competitive task accuracy.

Driverless simulation of path tracking based on PID control

In order to realize the path tracking of autonomous vehicles, this paper first simplifies the complex structure of the vehicle into a two-degree-of-freedom kinematic and dynamic vehicle model. based on PID control, the vehicle path tracking controller is designed to calculate the lateral deviation between the expected path and the actual path, and outputs the front wheel rotation angle through the controller, so that the autonomous vehicle can travel well according to the expected path. The control system model is built in the MATLAB/Simulink, and the vehicle path tracking control system is simulated and verified jointly with the CarSim software platform. Simulation results show that the designed control strategy can ensure the path tracking performance of intelligent vehicles at different speeds, and has good tracking accuracy, real-time performance and vehicle driving stability.

An Improved Kinematic Model Predictive Control for High-Speed Path Tracking of Autonomous Vehicles

Kinematic model predictive control (MPC) is well known for its simplicity and computational efficiency for path tracking of autonomous vehicles, however, it merely works well at low speed. In addition, earlier studies have demonstrated that tracking accuracy is improved by the feedback of yaw rate, as it improves the system transients. With this in mind, it is expected that the performance of path tracking can be improved by a cascaded controller that utilizes kinematic MPC to determine desired yaw rate rather than steering angle, and uses proportional-integral-derivative (PID) control to follow the reference yaw rate. However, directly combining MPC with PID feedback control of yaw rate results in a controller with poor tracking accuracy. The simulation results show that the cascaded MPC-PID controller has relatively stable but larger error compared to classic kinematic and dynamic MPC. Based on the analysis of vehicle sideslip angle, a novel path tracking control method is proposed, which is designed using a kinematic MPC to handle the disturbances on road curvature, a PID feedback control of yaw rate to reject uncertainties and modeling errors, and a vehicle sideslip angle compensator to correct the kinematic model prediction. The proposed controller performances involving steady-state and transient response, robustness, and computing efficiency were evaluated on Carsim/Matlab joint simulation environment. Furthermore, field experiments were conducted to validate the robustness against sensor disturbances and time lag. The results demonstrate that the developed vehicle sideslip compensator is sufficient to capture steer dynamics, and the developed controller significantly improves the performance of path tracking and follows the desired path very well, ranging from low speed to high speed even at the limits of handling.

No Blind Spots: Full-Surround Multi-Object Tracking for Autonomous Vehicles Using Cameras and LiDARs

Online multi-object tracking (MOT) is extremely important for high-level spatial reasoning and path planning for autonomous and highly-automated vehicles. In this paper, we present a modular framework for tracking multiple objects (vehicles), capable of accepting object proposals from different sensor modalities (vision and range) and a variable number of sensors, to produce continuous object tracks. This work is a generalization of the MDP framework for MOT proposed by Xiang et al. , with some key extensions - First, we track objects across multiple cameras and across different sensor modalities. This is done by fusing object proposals across sensors accurately and efficiently. Second, the objects of interest (targets) are tracked directly in the real world . This is a departure from traditional techniques where objects are simply tracked in the image plane. Doing so allows the tracks to be readily used by an autonomous agent for navigation and related tasks. To verify the effectiveness of our approach, we test it on real world highway data collected from a heavily sensorized testbed capable of capturing full-surround information. We demonstrate that our framework is well-suited to track objects through entire maneuvers around the ego-vehicle, some of which take more than a few minutes to complete. We also leverage the modularity of our approach by comparing the effects of including/excluding different sensors, changing the total number of sensors, and the quality of object proposals on the final tracking result.

Trajectory tracking for autonomous vehicles on varying road surfaces by friction-adaptive nonlinear model predictive control

ABSTRACTWe propose an adaptive nonlinear model predictive control (NMPC) for vehicle tracking control. The controller learns in real time a tyre force model to adapt to a varying road surface that is only indirectly observed from the effects of the tyre forces determining the vehicle dynamics. Learning the entire tyre model from data would require driving in the unstable region of the vehicle dynamics with a prediction model that has not yet converged. Instead, our approach combines NMPC with a noise-adaptive particle filter for vehicle state and tyre stiffness estimation and a pre-determined library of tyre models. The stiffness estimator determines the linear component of the tyre model during normal vehicle driving, and the control strategy exploits a relation between the tyre stiffness and the nonlinear part of the tyre force to select the appropriate full tyre model from the library, which is then used in the NMPC prediction model. We validate the approach in simulation using real vehicle parameters, demonstrate the real-time feasibility in automotive-grade processors using a rapid prototyping unit, and report preliminary results of experimental validation on a snow-covered test track.

Trajectory Tracking of Autonomous Vehicle with the Fusion of DYC and Longitudinal\u2013Lateral Control

The current research of autonomous vehicle motion control mainly focuses on trajectory tracking and velocity tracking. However, numerous studies deal with trajectory tracking and velocity tracking separately, and the yaw stability is seldom considered during trajectory tracking. In this research, a combination of the longitudinal–lateral control method with the yaw stability in the trajectory tracking for autonomous vehicles is studied. Based on the vehicle dynamics, considering the longitudinal and lateral motion of the vehicle, the velocity tracking and trajectory tracking problems can be attributed to the longitudinal and lateral control. A sliding mode variable structure control method is used in the longitudinal control. The total driving force is obtained from the velocity error in order to carry out velocity tracking. A linear time-varying model predictive control method is used in the lateral control to predict the required front wheel angle for trajectory tracking. Furthermore, a combined control framework is established to control the longitudinal and lateral motions and improve the reliability of the longitudinal and lateral direction control. On this basis, the driving force of a tire is allocated reasonably by using the direct yaw moment control, which ensures good yaw stability of the vehicle when tracking the trajectory. Simulation results indicate that the proposed control strategy is good in tracking the reference velocity and trajectory and improves the performance of the stability of the vehicle.

Kinematics-aware model predictive control for autonomous high-speed tracked vehicles under the off-road conditions

Abstract Although accurate trajectory tracking of autonomous vehicles strongly depends on a precise model, it is difficult to establish a precise model for high-speed tracked vehicles under off-road conditions, due to the complicated interactions between the tracks and the terrain. This paper presents a novel trajectory tracking methodology, called kinematics-aware model predictive control (KAMPC), by combining the slip kinematic model with a trajectory tracking control strategy for skid-steered tracked vehicles. For the slip kinematic model, an online identification methodology called six-parameter slip parameter estimation (SSPE) algorithm is proposed based on the instantaneous centers of rotation. For the trajectory tracking control strategy, optimized algorithm model predictive control is used to obtain the optimal control inputs. Finally, our method is validated through extensive experiments based on a distributed electric-drive high-speed tracked vehicle test platform over different types of terrain and varied off-road surface conditions. Experiment results show that the tracked vehicle kinematic characteristics are distinctive in off-road conditions, and the slipping and skidding that occur when turning are critical to this work. Compared with the other three widely used types of methodologies, the present KAMPC approach can reduce the average tracking residuals and has the best statistical performance.

Robust hierarchical controller with conditional integrator based on small gain theorem for reference trajectory tracking of autonomous vehicles

ABSTRACTA robust trajectory tracking controller is designed for autonomous vehicles based on a hierarchical architecture to make the autonomous vehicle track a given reference trajectory. The controller consists of two sub controllers: kinematic controller and dynamic controller. Based on the kinematics of tracking reference trajectory, a desired yaw rate is calculated by kinematic controller to make the lateral deviation global asymptotic stable. Then, steering wheel angle is calculated by a vehicle dynamic controller to make the vehicle yaw rate converge to the desired value and make the vehicle dynamic stable. Conditional integration method is used in the sub controllers. This method guarantees global asymptotic stability of tracking reference values and considers the uncertainty of parameters and constraints of desired yaw rate and actuators. Then based on small-gain theorem, the condition of the finite-gain L stability is given to the hierarchical controller to ensure the interconnected sub systems stable and prevent the amplification of system disturbance. Finally, the effectiveness and robustness of the controller are validated by real vehicle experiments.