Scenario-Transferable Semantic Graph Reasoning for Interaction-Aware Probabilistic Prediction

Hu, Yeping; Zhan, Wei; Tomizuka, Masayoshi

doi:10.48550/arxiv.2004.03053

Cited by 9 publications

(25 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such a representation is compact, efficient and generic, which captures sufficient information for intention determination and can be generically used across different driving scenarios. For more information related to DIA and SG, please refer to [18]. Illustrated in the left part of Figure 2, the process of extracting semantic graph representation G t−T f ,t from raw observations O t−T f ,t can also be formally described:…”

Section: Proposed Methodsmentioning

confidence: 99%

“…In this section, we first adopt the semantic graph (SG), which is originally introduced in [18]. In the SG, the dynamic insertion area (DIA) is defined as a generic and compact representation of the scenario.…”

Section: A High-level Intention-identification Policymentioning

confidence: 99%

“…Mathematically, we define each dynamic insertion area as With the extracted DIAs, the 3D spatial-temporal semantic graph G t−T h →t = (N t−T h →t , E t−T h →t ) can be constructed, where t − T h → t denotes the time span from a previous time step t − T h to current time step t with T h denoting the horizon. For more detailed description on the DIA properties, DIA extraction algorithm, feature selection, semantic graph construction, please refer to [18] [22].…”

Section: A1 Semantic Graphmentioning

confidence: 99%

“…As the high-level intention identification policy, the architecture of the SGN is shown in Figure 2. SGN takes the spatial-temporal 3D semantic graph from historic time step t − T h to current time step t as the input, rather than only spatial 2D semantic graph of current time step t in [18]. Such a change aims at capturing more temporal dynamics and interactions among vehicles.…”

Section: A2 Semantic Graph Networkmentioning

confidence: 99%

“…where G s denotes all the training graph samples; N s denotes all the nodes in one training graph sample; yi and wi denotes the ground-truth label for goal state and insertion probability of node n i . Though our goal is to predict ego vehicle's future motion, we output the goal state for all interacting vehicles rather than ego vehicle exclusively [18] to encourage sufficient relationship and interaction reasoning, and also realize data augmentation.…”

Section: A24 Loss Functionmentioning

confidence: 99%

See 4 more Smart Citations

Hierarchical Adaptable and Transferable Networks (HATN) for Driving Behavior Prediction

Wang¹,

Hu²,

Sun³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

When autonomous vehicles still struggle to solve challenging situations during on-road driving, humans have long mastered the essence of driving with efficient transferable and adaptable driving capability. By mimicking humans' cognition model and semantic understanding during driving, we present HATN, a hierarchical framework to generate high-quality driving behaviors in multi-agent dense-traffic environments. Our method hierarchically consists of a high-level intention identification and low-level action generation policy. With the semantic sub-task definition and generic state representation, the hierarchical framework is transferable across different driving scenarios. Besides, our model is also able to capture variations of driving behaviors among individuals and scenarios by an online adaptation module. We demonstrate our algorithms in the task of trajectory prediction for real traffic data at intersections and roundabouts, where we conducted extensive studies of the proposed method and demonstrated how our method outperformed other methods in terms of prediction accuracy and transferability.Preprint. Under review.

show abstract

Section: Proposed Methodsmentioning

confidence: 99%

Section: A High-level Intention-identification Policymentioning

confidence: 99%

Section: A1 Semantic Graphmentioning

confidence: 99%

Section: A2 Semantic Graph Networkmentioning

confidence: 99%

Section: A24 Loss Functionmentioning

confidence: 99%

See 3 more Smart Citations

Hierarchical Adaptable and Transferable Networks (HATN) for Driving Behavior Prediction

Wang¹,

Hu²,

Sun³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

A cognition‐inspired trajectory prediction method for vehicles in interactive scenarios

Xie

Wang

2023

IET Intelligent Trans Sys

View full text Add to dashboard Cite

Trajectory prediction of the ego vehicle is necessary for the cooperation driving of intelligent vehicles and drivers. Methods based on deep learning can fit complex functions, but they usually focus on vehicles' behavioral characteristics. However, vehicles' trajectories are closely related to the cognition results of drivers. Therefore, based on drivers' cognitive characteristics, a network model is designed to predict vehicle trajectories. Specifically, in the perception stage, featured grids are used that are in the driver's view to encode perceptual information; in the decision stage, convolution and graph attention operations are combined to model the driver's interaction with the surrounding traffic elements; in the motion stage, the elements are constrained in one hidden layer by vehicles' actual control inputs and design the corresponding method to obtain probabilistic results. With experiments in two typical scenarios, including intersection and roundabout, the proposed method can obtain reasonable prediction accuracy and generalizability. Meanwhile, abundant experiments are conducted and the results are compared, which reveal some common problems when predicting vehicle trajectories, particularly based on drivers' cognitive characteristics. These lessons learned from this study are summarized which may be useful for newcomers.

show abstract

To Develop Human-like Automated Driving Strategy Based on Cognitive Construction: Appraisal and Perspective

Xie

Chen

Tomizuka

et al. 2020

2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)

View full text Add to dashboard Cite

Scenario-Transferable Semantic Graph Reasoning for Interaction-Aware Probabilistic Prediction

Cited by 9 publications

References 45 publications

Hierarchical Adaptable and Transferable Networks (HATN) for Driving Behavior Prediction

Hierarchical Adaptable and Transferable Networks (HATN) for Driving Behavior Prediction

A cognition‐inspired trajectory prediction method for vehicles in interactive scenarios

To Develop Human-like Automated Driving Strategy Based on Cognitive Construction: Appraisal and Perspective

Contact Info

Product

Resources

About