CoverNet: Multimodal Behavior Prediction Using Trajectory Sets

Phan-Minh, Tung; Grigore, Elena Corina; Boulton, Freddy A.; Beijbom, Oscar; Wolff, Eric M.

doi:10.1109/cvpr42600.2020.01408

Cited by 330 publications

(236 citation statements)

References 20 publications

Supporting

Mentioning

235

Contrasting

Order By: Relevance

“…In urban scenarios oftentimes the high-level planning problem is solved with a discrete strategy space. One selects a trajectory from a finite candidate set [23], [32]. In that case, for each candidate trajectory one computes the associated costs and sorts them according to the lexicographic preference.…”

Section: B Optimization Over Lexicographic Preferencesmentioning

confidence: 99%

Urban Driving Games With Lexicographic Preferences and Socially Efficient Nash Equilibria

Zanardi

Mion

Bruschetta

et al. 2021

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

We describe Urban Driving Games (UDGs) as a particular class of differential games that model the interactions and incentives of the urban driving task. The drivers possess a "communal" interest, such as not colliding with each other, but are also self-interested in fulfilling traffic rules and personal objectives. Subject to their physical dynamics, the preference of the agents is expressed via a lexicographic relation that puts as first priority the shared objective of not colliding. Under mild assumptions, we show that communal UDGs have the structure of a lexicographic ordinal potential game which allows us to prove several interesting properties. Namely, socially efficient equilibria can be found by solving a single (lexicographic) optimal control problem and iterated best response schemes have desirable convergence guarantees.

show abstract

Section: B Optimization Over Lexicographic Preferencesmentioning

confidence: 99%

Urban Driving Games With Lexicographic Preferences and Socially Efficient Nash Equilibria

Zanardi

Mion

Bruschetta

et al. 2021

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

show abstract

“…Secondly, it predicts the offset of each grid unit through the fine scale of features. The CoverNet [8] uses the current and past states of the moving target to calculate the multi-modal probability distribution of the future state. It limits the possible future state set of the moving target within a reasonable prediction range, so as to achieve the maximum possible choice of the target's trajectory at the intersection.…”

Section: Related Workmentioning

confidence: 99%

“…It can not only solve the problems of data overflow and underflow, but also improve the speed and stability of calculation process. The proof process of the above theory is as shown in equation (8). M is the maximum value of {xi}.…”

Section: Normalization Functionmentioning

confidence: 99%

DeepPTP: A Deep Pedestrian Trajectory Prediction Model for Traffic Intersection

Lv¹,

Li²,

Dong³

et al. 2021

KSII TIIS

View full text Add to dashboard Cite

Compared with vehicle trajectories, pedestrian trajectories have stronger degrees of freedom and complexity, which poses a higher challenge to trajectory prediction tasks. This paper designs a mode to divide the trajectory of pedestrians at a traffic intersection, which converts the trajectory regression problem into a trajectory classification problem. This paper builds a deep model for pedestrian trajectory prediction at intersections for the task of pedestrian shortterm trajectory prediction. The model calculates the spatial correlation and temporal dependence of the trajectory. More importantly, it captures the interactive features among pedestrians through the Attention mechanism. In order to improve the training speed, the model is composed of pure convolutional networks. This design overcomes the single-step calculation mode of the traditional recurrent neural network. The experiment uses Vulnerable Road Users trajectory dataset for related modeling and evaluation work. Compared with the existing models of pedestrian trajectory prediction, the model proposed in this paper has advantages in terms of evaluation indicators, training speed and the number of model parameters.

show abstract

“…To solve these problems, various studies have recently attempted to predict trajectories using deep learning-based approaches [ 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 ] capable of learning data; consequently, it is not necessary to express the maneuver interaction mathematically. To consider the interactions efficiently, previous studies have attempted to model them by representing the information of vehicles as grid-shaped images [ 5 , 10 , 11 , 12 , 13 ] or graph structures [ 6 , 7 , 8 , 14 , 15 , 16 , 17 , 18 ] rather than using raw sensor data.…”

Section: Introductionmentioning

confidence: 99%

Vehicle Trajectory Prediction Using Hierarchical Graph Neural Network for Considering Interaction among Multimodal Maneuvers

Sunwoo²,

Lee³

2021

Sensors

View full text Add to dashboard Cite

Predicting the trajectories of surrounding vehicles by considering their interactions is an essential ability for the functioning of autonomous vehicles. The subsequent movement of a vehicle is decided based on the multiple maneuvers of surrounding vehicles. Therefore, to predict the trajectories of surrounding vehicles, interactions among multiple maneuvers should be considered. Recent research has taken into account interactions that are difficult to express mathematically using data-driven deep learning methods. However, previous studies have only considered the interactions among observed trajectories due to subsequent maneuvers that are unobservable and numerous maneuver combinations. Thus, to consider the interaction among multiple maneuvers, this paper proposes a hierarchical graph neural network. The proposed hierarchical model approximately predicts the multiple maneuvers of vehicles and considers the interaction among the maneuvers by representing their relationships in a graph structure. The proposed method was evaluated using a publicly available dataset and a real driving dataset. Compared with previous methods, the results of the proposed method exhibited better prediction performance in highly interactive situations.

show abstract

CoverNet: Multimodal Behavior Prediction Using Trajectory Sets

Cited by 330 publications

References 20 publications

Urban Driving Games With Lexicographic Preferences and Socially Efficient Nash Equilibria

Urban Driving Games With Lexicographic Preferences and Socially Efficient Nash Equilibria

DeepPTP: A Deep Pedestrian Trajectory Prediction Model for Traffic Intersection

Vehicle Trajectory Prediction Using Hierarchical Graph Neural Network for Considering Interaction among Multimodal Maneuvers

Contact Info

Product

Resources

About