Embedded Files
Social-aware and interactive decision-making of autonomous vehicles (AVs)
Decision-making in driving is a complex and significant challenge, especially when ensuring AVs’ interactive and socially compatible behavior within mixed-traffic environments. The intricacy arises from the frequent interactions among surrounding vehicles and the often partial or complete unknowns about their actions and intentions. This complexity underscores the need for robust decision-making strategies to operate effectively even when information is unavailable.
Safety-critical autonomous vehicle control
Safety-critical autonomous vehicle control
Safety is a paramount concern in the automotive sector, serving as a linchpin for both successful deployment and consumer confidence. While features such as adaptive cruise control (ACC) and lane-keeping assist (LKS) effectively handle straightforward driving situations, the complexity escalates when AV systems encounter nuanced scenarios, such as LCs and navigating intersections. Controllers must skillfully navigate potential hazards within these intricate contexts, ensuring prompt and efficient risk mitigation. AV control algorithms, facing the challenge of managing various constraints, safety-critical conditions, adherence to traffic regulations, and decisions made by higher-level behavior planners and local trajectory planners, are at the forefront of this complexity. These algorithms must integrate these diverse components seamlessly to ensure AVs’ safe and smooth operation in real-world environments.
Connected and automated platoon control
Connected and automated platoon control
In the coming years, intelligent transportation systems will address traffic and energy inefficiencies through advanced control algorithms and wireless communication, revolutionizing smart cities with AVs and modern road infrastructure. Automated platooning, where AVs communicate via vehicle-to-vehicle (V2V) technology to maintain optimal spacing and speed, will significantly reduce congestion, instability, and fuel waste and improve safety. Effective control systems for platoons must ensure individual and string stability, meaning that vehicle spacing errors do not propagate down the string, depending on information flow topologies (IFTs) and spacing policies.
Longitudinal and Lateral Control with Deep Reinforcement Learning
In the realm of intelligent control, reinforcement learning (RL) has recently shown significant promise. Researchers have developed a novel deep reinforcement learning-based approach to optimal vehicle control by integrating deep learning into the RL framework. This is termed as deep reinforcement learning (DRL). Moreover, we can explore the application of multi-agent deep reinforcement learning (MADRL) for advanced vehicle control problems. By assigning specific control tasks to individual agents — such as acceleration, braking, steering, and lane-keeping — we create a collaborative environment in which agents learn to optimize their respective functions. This distributed approach enhances the system’s ability to handle complex, dynamic driving scenarios, improve real-time decision-making, and adapt to changing conditions.
Energy-aware and Coordination Platform for Advanced Air Mobility (AAM)
The central objective is to establish a robust analytical and operational framework that connects flight energy consumption, battery optimization, grid impact modeling, and vertiport siting strategies into a cohesive decision-support structure.
Flight Energy Consumption Modeling. This component involves constructing high-fidelity models that predict eVTOL energy use under various mission profiles, environmental conditions, and payload configurations. These models quantify consumption across flight phases—takeoff, cruise, and landing—to evaluate how flight dynamics and urban airspace constraints influence overall energy demand.
Battery Control and Optimization
The emphasis lies on developing algorithms that manage and extend the lifecycle of onboard energy storage systems. By incorporating thermal dynamics, degradation effects, and power density constraints, these models allow control strategies that ensure optimal charging, discharging, and energy use, aligning with mission energy needs and grid constraints.
Page updated
Google Sites
Report abuse