A5103002075- Autonomous Vehicle Technology and Safety
- Traffic control and management
- Traffic Prediction and Management Techniques
- Advanced Measurement and Detection Methods
- Soil Geostatistics and Mapping
- Traffic and Road Safety
- Soil Moisture and Remote Sensing
- Advanced Control Systems Optimization
- Computer Graphics and Visualization Techniques
- Maritime Navigation and Safety
- Software Engineering Research
- Statistical and Computational Modeling
- Reservoir Engineering and Simulation Methods
- Capital Investment and Risk Analysis
- Fault Detection and Control Systems
- Video Surveillance and Tracking Methods
- Mining Techniques and Economics
- Imbalanced Data Classification Techniques
- Advanced Vision and Imaging
- Human Motion and Animation
- Maritime Ports and Logistics
- Maritime Security and History
- Optical Systems and Laser Technology
- Bayesian Modeling and Causal Inference
- Soil and Unsaturated Flow
Taiyuan University of Technology
2024
Jilin University
2023-2024
Ocean University of China
2024
State Key Laboratory on Integrated Optoelectronics
2024
We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as preliminary step, demonstrates remarkable capabilities. Through RL, naturally emerges with numerous powerful intriguing behaviors. However, it encounters challenges such poor readability, language mixing. To address these issues further enhance performance, we DeepSeek-R1, which incorporates...
Car-following behavior is one of the most common driving behaviors. To reduce impact driver reaction delay and accumulated errors in predicting long sequences on accuracy speed prediction, we propose a deep learning car-following model based sequence-to-sequence (seq2seq) architecture with an attention mechanism. Firstly, analyze characteristics during process design mechanism to learn probability distribution delay. This allows consider more environmental information at moment when actually...
The complexity inherent in navigating intricate traffic environments poses substantial hurdles for intelligent driving technology. continual progress mapping and sensor technologies has equipped vehicles with the capability to intricately perceive their exact position interplay among surrounding elements. Building upon this foundation, paper introduces a deep reinforcement learning method solve decision-making trajectory planning problem of vehicles. employs framework feature extraction,...
Generating rich and controllable motion is a pivotal challenge in video synthesis. We propose Boximator, new approach for fine-grained control. Boximator introduces two constraint types: hard box soft box. Users select objects the conditional frame using boxes then use either type of to roughly or rigorously define object's position, shape, path future frames. functions as plug-in existing diffusion models. Its training process preserves base model's knowledge by freezing original weights...
Data-driven predictive control promises modelfree wave-dampening strategies for Connected and Autonomous Vehicles (CAVs) in mixed traffic flow. However, the performance suffers from unknown noise disturbances, which could occur offline data collection online control. In this paper, we propose a Robust Data-EnablEd Predictive Leading Cruise Control (RDeeP-LCC) method based on reachability analysis, aiming to achieve safe optimal of CAVs under bounded process external disturbances. Precisely,...
This paper presents an enhanced ground vehicle localization method designed to address the challenges associated with state estimation for autonomous vehicles operating in diverse environments. The focus is specifically on precise of position and orientation both local global coordinate systems. proposed approach integrates estimates generated by existing visual-inertial odometry (VIO) methods into information obtained from Global Navigation Satellite System (GNSS). integration achieved...
Controlling mixed platoons, which consist of both connected and automated vehicles (CAVs) human-driven (HDVs), poses significant challenges due to the uncertain unknown human driving behaviors. Data-driven control methods offer promising solutions by leveraging available trajectory data, but their performance can be compromised process noise adversarial attacks. To address this issue, paper proposes a Robust Data-EnablEd Predictive Leading Cruise Control (RDeeP-LCC) framework based on...
Free lane change (FLC) is an important research direction of intelligent driving vehicles. In this paper, a decision model based on deep learning established with ego and adjacent risk properties environmental A long short-term memory neural network that can be associated time series characteristics used to the process analysis factors affecting decision. The result decision-making training shows recognition accuracy keeping reaches more than 92%. Then, human-like FLC trajectory planned...
In the domain of intelligent driving vehicles, radar and camera sensors play a crucial role, their fusion has been widely acknowledged to enhance sensing range accuracy. This article delves into challenge asynchronous information for object detection between camera, taking consideration different sampling frequencies camera. Initially, state observation prediction models are formulated, leading proposal comprehensive target tracking model. Subsequently, matching process results is...
In a dynamic environment, autonomous driving vehicles require accurate decision-making and trajectory planning. To achieve this, need to understand their surrounding environment predict the behavior future trajectories of other traffic participants. recent years, vectorization methods have dominated field motion prediction due ability capture complex interactions in scenes. However, existing research using for scene encoding often overlooks important physical information about vehicles, such...