Selected Published Papers
Hierarchical Trajectory Planning Based on Adaptive Motion Primitives and Bilevel Corridor
with Wenshuo Wang, Boyang Wang, Haijie Guan, Haiou Liu, Shaobin Wu, and Huiyan Chen.
IEEE Transactions on Vehicular Technology
Abstract: This paper presents an efficient and risk-aware search-optimization hierarchical trajectory planning method for automated vehicles in different road structure. The proposed approach incorporates a time-series motion risk field, capturing diverse road structures through a spatiotemporal map. Then, an adaptive motion primitive is developed, dynamically adjusting action time windows based on evolving risk and expected deviation during the search process. This enables efficient and accurate initial trajectory generation. Additionally, a bilevel corridor is introduced to extract the drivable area and rerepresent the risk field, enabling trajectory smoothing to consider motion risk without resorting to non-convex optimization methods. The method is validated through simulation in structured and unstructured scenarios, demonstrating improved efficiency, flexibility, and optimization quality compared to fixed-step search and single-level corridor-based optimization approaches.
Multi-Vehicle Collaborative Lane Changing Based on Multi-Agent Reinforcement Learning
with Xiang Zhang, Boyang Wang, Mingxuan Xue, Zhiwei Li, Haiou Liu
2024 IEEE Intelligent Vehicles Symposium (IV)
Abstract: Achieving safe lane changing is a crucial function of autonomous vehicles, with the complexity and uncertainty of interaction involved. Learning-based approaches and vehicle collaboration techniques can enhance vehicles’ awareness of the dynamic environment, thereby enhancing the interactive capabilities. Therefore, this paper proposes a Multi-Agent Reinforcement Learning (MARL) approach to coordinate connected vehicles in reaching their respective lane changing targets. Vehicle state, scene elements, potential risk, and intention information are abstracted into highly expressive vectorized inputs. Based on this, a lightweight parameter-sharing network framework is designed to learn safe and robust cooperative lane changing policies. To address the challenges arising from multi-objects and multi-targets, a Prioritized Action Extrapolation (PAE) mechanism is employed to train the network. Through priority assignment and action extrapolation, the proposed MARL approach can optimize the decision sequence dynamically and enhance the interaction in multi-vehicle scenarios, thereby improving the vehicles’ intention attainment rate. Simulated experiments in 2-lane and 3-lane scenarios have been conducted to verify the adaptability and performance of the proposed MARL method.
Modeling and Quantitative Evaluation Method of Environmental Complexity for Measuring Autonomous Capabilities for Unmanned Ground Vehicles
with Shaobin Wu, Jianwei Gong, Zexin Yan
Unmanned Systems
Abstract: This paper proposes a sampling-based multi-dimensional entropy hierarchical evaluation method to evaluate the environmental complexity for measuring autonomous capabilities of unmanned ground vehicles. Through establishing the multi-dimensional environment model, the complexity of environmental elements in various dimensions is measured by combining the analytic hierarchy process and the improved gravitational field model. Based on the graph entropy and the environment segmentation sampling strategy, the environmental complexity is comprehensively evaluated from the two perspectives of the objective complexity of the environmental structure and the subjective complexity of environmental characteristics. The evaluation of the actual test environment shows that the environmental complexity evaluation model can effectively reflect the individual complexity differences of environmental elements, and achieve the comprehensive complexity evaluation of the environment including multiple test scenarios, which provides a basis for the test scenario design and measuring autonomous capabilities of unmanned ground vehicles.
Working Papers
Optimization-Based Spatiotemporal Trajectory Planning with Behavior Cells for Autonomous Driving
with Wenshuo Wang, Zhide Zhang, Xin Xia, Boyang Wang, Chao Lu and Haiou Liu.
IEEE Transactions on Intelligent Transportation Systems (under review)
Abstract: Achieving safe and executable trajectory planning in traffic scenarios necessitates considering vital environmental elements, ego behaviors, and vehicle inematics, which poses challenges of real-time performance and convergence for optimization-based methods. To address these issues, we propose an optimization-based trajectory planning framework that handles environmental elements and provides heuristic behavior constraints at the spatiotemporal drivable domain level, followed by progressively optimization to achieve fast and stable convergence. Specifically, considering vehicle motion capabilities and the tactical decisions for various environmental elements, we first partition the spatiotemporal domain into modular drivable cells with behavior semantics, refer to Behavior Cells (BCs). All feasible candidate behaviors are then rapidly enumerated through BCs combinations. ubsequently, the Finite-Horizon Markov Decision Process (FHMDP) is utilized to model and evaluate each behavior in the spatiotemporal drivable domain, with the optimal BCs combination serving as heuristic constraints. Finally, we utilize a dynamic two-stage optimization approach to progressively satisfy the planning-related requirements, generating trajectories within the selected BCs combination with fast convergence. Simulation in various typical traffic demonstrate that the proposed method showcases stable reliability and realtime performance compared to the baselines under different traffic densities. On-road experiments further validate the effectiveness of the proposed method in real-world traffic scenarios. Website: https://lshasd123.github.io/Behavior-Cells/.
Work in Progress
PHISI: Suspension-aware Vehicle-Terrain Contact model for Pose Estimation in Unstructured Scenarios
with Ji Li.
Abstract:
Patents
-
A method for safety verification and control in unmanned off-road vehicles navigating through obstacle terrain. NO. 202311088086.3, China.
-
A method for safety behavior detection in autonomous off-road vehicles. NO. 202311088074.0, China.
-
A method and system for evaluating the quality of unmanned vehicle path planning. NO. 202221836601.2, China.
-
A convective heat exchange device based on vortex low-speed circulation air supply. NO. 201910247180.6, China.