Search Results

SLAM (Simultaneous Localization and Mapping) plays a key role in autonomous driving. Recently, 4D Radar has attracted widespread attention because it breaks through the limitations of 3D millimeter wave radar and can simultaneously detect the distance, velocity, horizontal azimuth and elevation azimuth of the target with high resolution. However, there are few studies on 4D Radar in SLAM. In this paper, RI-FGO, a 4D Radar-Inertial SLAM method based on Factor Graph Optimization, is proposed. The RANSAC (Random Sample Consensus) method is used to eliminate the dynamic obstacle points from a single scan, and the ego-motion velocity is estimated from the static point cloud. A 4D Radar velocity factor is constructed in GTSAM to receive the estimated velocity in a single scan as a measurement and directly integrated into the factor graph. The 4D Radar point clouds of consecutive frames are matched as the odometry factor.

Speech enhancement can extract clean speech from noise interference, enhancing its perceptual quality and intelligibility. This technology has significant applications in in-car intelligent voice interaction. However, the complex noise environment inside the vehicle, especially the human voice interference is very prominent, which brings great challenges to the vehicle speech interaction system. In this paper, we propose a speech enhancement method based on target speech features, which can better extract clean speech and improve the perceptual quality and intelligibility of enhanced speech in the environment of human noise interference. To this end, we propose a design method for the middle layer of the U-Net architecture based on Long Short-Term Memory (LSTM), which can automatically extract the target speech features that are highly distinguishable from the noise signal and human voice interference features in noisy speech, and realize the targeted extraction of clean speech.

The on-ramp merging driving scenario is challenging for achieving the highest-level autonomous driving. Current research using reinforcement learning methods to address the on-ramp merging problem of automated vehicles (AVs) is mainly designed for a single AV, treating other vehicles as part of the environment. This paper proposes a control framework for cooperative on-ramp merging of multiple AVs based on multi-agent deep reinforcement learning (MADRL). This framework facilitates AVs on the ramp and adjacent mainline to learn a coordinate control policy for their longitudinal and lateral motions based on the environment observations. Unlike the hierarchical architecture, this paper integrates decision and control into a unified optimal control problem to solve an on-ramp merging strategy through MADRL.

This paper presents an integrated modeling approach for real-time discretionary lane-changing decisions by autonomous vehicles, aiming to achieve human-like behavior. The approach incorporates a two-player normal-form game and a novel risk field method. The normal-form game represents the strategic interactions among traffic participants. It captures the trade-offs between lane-changing benefits and risks based on vehicle motion states during a lane change. By continuously determining the Nash equilibrium of the game at each time step, the model decides when it is appropriate to change the lane. A novel risk field method is integrated with the game to model risks in the game pay-offs. The risk field introduces regions along the desired target lane with different time headway ranges and risk weights, capturing traffic participants' complex risk perceptions and considerations in lane-changing scenarios.

Lane change obstacle avoidance is a common driving scenario for autonomous vehicles. However, existing methods for lane change obstacle avoidance in vehicles decouple path and velocity planning, neglecting the coupling relationship between the path and velocity. Additionally, these methods often do not sufficiently consider the lane change behaviors characteristic of human drivers. In response to these challenges, this paper innovatively applies the Dynamic Movement Primitives (DMPs) algorithm to vehicle trajectory planning and proposes a real-time trajectory planning method that integrates DMPs and Artificial Potential Fields (APFs) algorithm (DMP-Fs) for lane change obstacle avoidance, enabling rapid coordinated planning of both path and velocity. The DMPs algorithm is based on the lane change trajectories of human drivers. Therefore, this paper first collected lane change trajectory samples from on-road vehicle experiments.

In the field of vehicle aerodynamic simulation, Reynold Averaged Navier-Stokes (RANS) model is widely used due to its high efficiency. However, it has some limitations in capturing complex flow features and simulating large separated flows. In order to improve the computational accuracy within a suitable cost, the Field Inversion and Machine Learning (FIML) method, based on a data-driven approach, has received increasing attention in recent years. In this paper, the optimal coefficients of the Generalized k-ω (GEKO) model are firstly obtained by the discrete adjoint method of FIML, utilizing the results of wind tunnel experiments. Then, the mapping relationship between the flow field characteristics and the optimal coefficients is established by a neural network to augment the turbulence model.

Recently, several datasets have become available for occupant monitoring algorithm development, including real and synthetic datasets. However, real data acquisition is expensive and labeling is complex, while virtual data may not accurately reflect actual human physiology. To address these issues and obtain high-fidelity data for training intelligent driving monitoring systems, we have constructed a hybrid dataset that combines real driving image data with corresponding virtual data generated from 3D driving scenarios. We have also taken into account individual anthropometric measures and driving postures. Our approach not only greatly enriches the dataset by using virtual data to augment the sample size, but it also saves the need for extensive annotation efforts. Besides, we can enhance the authenticity of the virtual data by applying ergonomics techniques based on RAMSIS, which is crucial in dataset construction.

Accurate and reliable localization in GNSS-denied environments is critical for autonomous driving. Nevertheless, LiDAR-based and camera-based methods are easily affected by adverse weather conditions such as rain, snow, and fog. The 4D Radar with all-weather performance and high resolution has attracted more interest. Currently, there are few localization algorithms based on 4D Radar, so there is an urgent need to develop reliable and accurate positioning solutions. This paper introduces RIO-Vehicle, a novel tightly coupled 4D Radar/IMU/vehicle dynamics within the factor graph framework. RIO-Vehicle aims to achieve reliable and accurate vehicle state estimation, encompassing position, velocity, and attitude. To enhance the accuracy of relative constraints, we introduce a new integrated IMU/Dynamics pre-integration model that combines a 2D vehicle dynamics model with a 3D kinematics model.

LiDAR and camera fusion have emerged as a promising approach for improving place recognition in robotics and autonomous vehicles. However, most existing approaches often treat sensors separately, overlooking the potential benefits of correlation between them. In this paper, we propose a Cross- Modality Module (CMM) to leverage the potential correlation of LiDAR and camera features for place recognition. Besides, to fully exploit potential of each modality, we propose a Local-Global Fusion Module to supplement global coarse-grained features with local fine-grained features. The experiment results on public datasets demonstrate that our approach effectively improves the average recall by 2.3%, reaching 98.7%, compared with simply stacking of LiDAR and camera.

Radar is playing more and important role in multiple object detection and tracking system due to the fact that Radar can not only determine the velocity instantly but also it is less influenced by environment conditions. However, Radar faces the problem that it has many detection clutter，false alarms and detection results are easily affected by the reflected echoes of road boundary in traffic scenes. Besides this, With the increase of the number of targets and the number of effective echoes, the number of interconnection matrices increases exponentially in joint probability data association, which will seriously affect the real-time and accuracy of high-speed scene algorithms.in the tracking system. So, A method of using millimeter wave radar to detect and fit the boundary guardrail of high-speed road is proposed, and the fitting results are applied to the vehicle detection and tracking system to improve the tracking accuracy.

On account of the insufficient lane-changing scenario test cases and the inability to conduct graded evaluation testing in current autonomous driving system field testing, this paper proposed an approach that combined data-driven and knowledge-driven methods to extract lane-changing test concrete scenarios with graded risk levels for field testing. Firstly, an analysis of the potentially hazardous areas in lane-changing scenarios was conducted to derive key functional lane-changing scenarios. Three typical key functional lane-changing scenarios were selected, namely, lane-changing with a preceding vehicle braking, lane-changing with a preceding vehicle in the same direction, and lane-changing with a rear cruising vehicle in the adjacent lane, and their corresponding safety goals were respectively analyzed. Secondly, the GAMAB criterion was introduced as an evaluation standard for autonomous driving systems.

Positioning system is a key module of autonomous driving. As for LiDAR SLAM system, it faces great challenges in scenarios where there are repetitive and sparse features. Without loop closure or measurements from other sensors, odometry match errors or accumulated errors cannot be corrected. This paper proposes a construction method of LiDAR anchor constraints to improve the robustness of the SLAM system in the above challenging environment. We propose a robust anchor extraction method that adaptively extracts suitable cylindrical anchors in the environment, such as tree trunks, light poles, etc. Skewed tree trunks are detected by feature differences between laser lines. Boundary points on cylinders are removed to avoid misleading. After the appropriate anchors are detected, a factor graph-based anchor constraint construction method is designed. Where direct scans are made to anchor, direct constraints are constructed.

In order to improve the efficiency of safety performance test for intelligent vehicles and construct the test case set quickly, critical scenarios based on graded hazard disposal model of human drivers are proposed, which can be used for extraction of test cases for safety performance. Based on the natural driving data in China Field Operational Test (China-FOT), the four-stage collision avoidance process of human drivers is obtained, including steady driving stage, risk judgment stage, collision reaction stage and collision avoidance stage. And there are two human driver states: general state and alert state. Then the graded hazard disposal model of human drivers is constructed.

Reinforcement learning is a promising approach to solve the energy management for hybrid electric vehicles. In this paper, based on the DQN (Deep Q-Network) reinforcement learning algorithm which is widely used at present, double DQN, dueling DQN and learning from demonstration are integrated; states, actions, rewards and the experience pool based on the characteristics of series-parallel multi-speed hybrid powertrain are designed; the hybrid energy management strategy based on D4QN (Double Dueling Deep Q-Network with Demonstrations) algorithm is established. Based on the training results of D4QN algorithm, multi-parameter analysis under state and action space, HCU (Hybrid control unit) application and MIL (Model in-loop) test research are conducted.

The driver monitoring system (DMS) plays an essential role in reducing traffic accidents caused by human errors due to driver distraction and fatigue. The vision-based DMS has been the most widely used because of its advantages of non-contact and high recognition accuracy. However, the traditional RGB camera-based DMS has poor recognition accuracy under complex lighting conditions, while the IR-based DMS has a high cost. In order to improve the recognition accuracy of conventional RGB camera-based DMS under complicated illumination conditions, this paper proposes a lightweight low-illumination image enhancement network inspired by the Retinex theory. The lightweight aspect of the network structure is realized by introducing a pixel-wise adjustment function. In addition, the optimization bottleneck problem is solved by introducing the shortcut mechanism.

To improve the maneuverability and energy consumption of an electrical vehicle, a two-level speed control method based on model predictive control (MPC) is proposed for accurate control of the vehicle during downhill coasting. The targeted acceleration is planned using the anti-interference speed filter and MPC algorithm in the upper-level controller and executed using the integrated algorithm with the inverse vehicle dynamics and proportional-integral-derivative control model (PID) in the lower-level controller, improving the algorithm’s anti-interference performance and road adaptability. Simulations and vehicle road tests showed that the proposed method could realize accurate real-time speed control of the vehicle during downhill coasting. It can also achieve a smaller derivation between the actual and targeted speeds, as well as more stable speeds when the road resistance changes abruptly, compared with the conventional PID method.

Harmony-with-traffic refers to the ability of autonomous vehicles to maximize the driving benefits such as comfort, efficiency, and energy consumption of themselves and the surrounding traffic during interactive driving under traffic rules. In the test of harmony-with-traffic, one or more background vehicles that can respond to the driving behavior of the vehicle under test are required. For this purpose, the functional requirements of car-following model for harmony-with-traffic evaluation are analyzed from the dimensions of test conditions, constraints, steady state and dynamic response. Based on them, an interactive car-following model (ICFM) is developed. In this model, the concept of equivalent distance is proposed to transfer lateral influence to longitudinal. The calculation methods of expected speed are designed according to the different car-following modes divided by interaction object, reaction distance and equivalent distance.

The intersection is recognized as the most dangerous area because of the restricted road structures and indeterminate traffic regulations. Therefore, according to the Vehicle-to-everything (V2X) communication, Intelligent Transportation Systems (ITS), and Digital Twin data, we present a potential energy field method to establish the general characteristics of intersection traffic safety, evaluate the safety situation of intersection and assist intersection traffic participants in passing through the intersection safer and more efficient. The resulting potential energy field method is established by the contour line of traffic participants' potential energy, which is constructed as a superposition of disparate energies, such as boundary potential energy, body potential energy, and velocity potential energy. The intersection traffic safety is evaluated by the potential energy field characteristic of simultaneous intersection traffic participants.

With the popularity of automated vehicles, the future mixed traffic flow contains automated vehicles with different degrees of intelligence developed by other manufacturers. Therefore, simulating the interaction behavior of automated vehicles with varying levels of intelligence is crucial for testing and evaluating autonomous driving systems. Since the algorithm of traffic vehicles with various intelligence levels is difficult to obtain, it leads to hardships in quantitatively characterizing their interaction behaviors. Therefore, this paper designs a new automated vehicle test platform to solve the problem. The intelligent vehicle testbed with multiple personalized in-vehicle control units in the loop consists of three parts: 1. Multiple controllers in the loop to simulate the behavior of traffic vehicles；2. The central console applies digital twin technology to share the same traffic scenario between the tested vehicle and the traffic vehicle, creating a mixed traffic flow. 3.