Simultaneous trajectory prediction for multiple heterogeneous traffic participants is essential safe and efficient operation of connected automated vehicles under complex driving situations. Two main challenges this task are to handle the varying number target agents jointly consider factors that would affect their future motions. This because different kinds have motion patterns, behaviors affected by individual dynamics, interactions with surrounding agents, as well infrastructures. A...

Predicting the behaviors of other agents on road is critical for autonomous driving to ensure safety and efficiency. However, challenging part how represent social interactions between output different possible trajectories with interpretability. In this paper, we introduce a neural prediction framework based Transformer structure model relationship among interacting extract attention target agent map waypoints. Specifically, organize into graph utilize multi-head encoder relations them. To...

Autonomous driving systems require a comprehensive understanding and accurate prediction of the surrounding environment to facilitate informed decision-making in complex scenarios. Recent advances learning-based have highlighted importance integrating planning. However, this integration poses significant alignment challenges through consistency between patterns, interaction future To address these challenges, we introduce Hybrid-Prediction integrated Planning (HPP) framework, which operates...

Grounding natural language in images, such as localizing "the black dog on the left of tree", is one core problems artificial intelligence, it needs to comprehend fine-grained and compositional space. However, existing solutions rely association between holistic features visual features, while neglect nature reasoning implied language. In this paper, we propose a grounding model that can automatically compose binary tree structure for parsing then perform along bottom-up fashion. We call our...

Connected automated driving technologies have shown tremendous improvement in recent years. However, it is still not clear how behaviors and energy consumption correlate with each other to what extent these factors related connected vehicles can influence the motion prediction performance. The precise recognition of vehicle critical safety for (CAVs). Hence, this study, an energy-aware pattern analysis system are proposed CAVs using a deep learning-based time-series modeling approach. First,...

Decision-making for urban autonomous driving is challenging due to the stochastic nature of interactive traffic participants and complexity road structures. Although reinforcement learning (RL)-based decision-making schemes are promising handle scenarios, they suffer from low sample efficiency poor adaptability. In this paper, we propose Scene-Rep Transformer enhance RL capabilities through improved scene representation encoding sequential predictive latent distillation. Specifically, a...

Predicting the future trajectory of a surrounding vehicle in congested traffic is one necessary abilities an autonomous vehicle. In congestion, vehicle's movement result its interaction with vehicles. A congestion may have many neighbors relatively short distance, while only small part affect mostly. this work, An interaction-aware method that predicts ego considering eight vehicles proposed. The dynamics are encoded by LSTMs shared weights, and extracted simple CNN. proposed model trained...

Integrating trajectory prediction to the decision-making and planning modules of modular autonomous driving systems is expected improve safety efficiency self-driving vehicles. However, a vehicle's future challenging task since it affected by social interactive behaviors neighboring vehicles, number vehicles can vary in different situations. This work proposes GNN-RNN based Encoder-Decoder network for interaction-aware prediction, where vehicles' dynamics features are extracted from their...

Predicting the future trajectory of surrounding vehicles is essential for navigation autonomous in complex real-world driving scenarios. It challenging as a vehicle's motion affected by many factors, including its infrastructures and vehicles. In this work, we develop ReCoG (Recurrent Convolutional Graph Neural Networks), which general scheme that represents vehicle interactions with infrastructure information heterogeneous graph applies neural networks (GNNs) to model high-level prediction....

Consensus of a network with directed acyclic graph, graph no cycles, is always guaranteed if it contains spanning tree. This paper studies the effect adding edges to that may result in cycle. It shown on consensus performance whole only determined by local subnetwork containing all added edges. More specifically, both one-dimensional (1-D) chain and 2-D grid are investigated this paper. proved that, when reverse edge added, degraded amount range, is, independent size or location edge.

An intriguing visible-light-induced strategy has been established for the P-H insertion reaction between acylsilanes and H-phosphorus oxides that, upon a subsequent acidic process, deliver wide variety of α-hydroxyphosphorus in good yields (up to 93% yield). The metal-free protocol represents unique example C-P bond formation through situ generation siloxycarbenes. This methodology features advantages operational simplicity, mild conditions, broad substrate scope, column free gram-scale synthesis.

Albeit notable endeavors in enantioselective carbene insertion into X–H bonds (X = C, O, N, S, Si, B), the catalytic asymmetric P–H reactions still stand for a long-lasting challenge. By merging transition-metal catalysis with organocatalysis, we achieve scalable transformation between diazo pyrazoleamides and H-phosphine oxides that upon subsequent reduction delivers wide variety of optically active β-hydroxyl phosphine good yields high enantioselectivity. The achiral copper catalyst...

Predicting the multimodal future motions of neighboring agents is essential for an autonomous vehicle to navigate complex scenarios.It challenging as motion agent affected by interaction among itself, other agents, and local roads.Unlike most existing works, which predict a fixed number possible agent, we propose map-adaptive predictor that can variable trajectories according lanes with candidate centerlines (CCLs).The predicts not only guided single CCLs but also scene-reasoning prediction...

This paper proposes a novel deep learning framework for multi-modal motion prediction. The consists of three parts: recurrent neural network to process target agent's process, convolutional the rasterized environment representation, and distance-based attention mechanism interactions among different agents. We validate proposed on large-scale real-world driving dataset, Waymo open compare its performance against other methods standard testing benchmark. qualitative results manifest that...

In this paper, we address the problem of secure pose estimation an autonomous vehicle (AV) under cyber attacks. An extended Kalman filter (EKF) is used to fuse measurements from multiple sensors including GPS, LIDAR, and IMU. To deal with possible sensor attacks, design a cumulative sum (CUSUM) detector monitor inconsistency between predicted via mathematical model measurement. EKF reconfiguration scheme proposed mitigate influence attacks once compromised identified. The feasibility...