Learning to Navigate Intersections with Unsupervised Driver Trait Inference

  title={Learning to Navigate Intersections with Unsupervised Driver Trait Inference},
  author={Shuijing Liu and Peixin Chang and Haonan Chen and Neeloy Chakraborty and Katherine Driggs Campbell},
  journal={2022 International Conference on Robotics and Automation (ICRA)},
Navigation through uncontrolled intersections is one of the key challenges for autonomous vehicles. Identifying the subtle differences in hidden traits of other drivers can bring significant benefits when navigating in such environments. We propose an unsupervised method for inferring driver traits such as driving styles from observed vehicle trajectories. We use a variational autoencoder with recurrent neural networks to learn a latent representation of traits without any ground truth trait… 

Figures and Tables from this paper



Reinforcement Learning for Autonomous Driving with Latent State Inference and Spatial-Temporal Relationships

This work encodes prior knowledge on the latent states of other drivers through a framework that combines the reinforcement learner with a supervised learner and model the influence passing between different vehicles through graph neural networks (GNNs).

Simultaneous policy learning and latent state inference for imitating driver behavior

This work proposes a method for learning driver models that can account for unobserved states and shows that latent state driving policies outperform baseline methods at replicating driver behavior.

Enhanced intelligent driver model to access the impact of driving strategies on traffic capacity

A new car-following model is proposed that also serves as the basis of an ACC implementation in real cars and eliminates the sometimes unrealistic behaviour of the IDM in cut-in situations with ensuing small gaps that regularly are caused by lane changes of other vehicles in dense or congested traffic.

A Study on Distance Measurement Module for Driving Vehicle Velocity Estimation in Multi-Lanes Using Drones

A method of estimating driving vehicle information usually uses a speed gun and a fixed speed camera. Estimating vehicle information using the speed gun has a high risk of traffic accidents by the

A Taxonomy and Review of Algorithms for Modeling and Predicting Human Driver Behavior.

A mathematical framework for describing the dynamics of interactive multi-agent traffic based on the partially observable stochastic game is introduced and provides a basis for discussing different driver modeling techniques.

Social NCE: Contrastive Learning of Socially-aware Motion Representations

This work intro-duce a social contrastive loss that regularizes the extracted motion representation by discerning the ground-truth positive events from synthetic negative ones, and constructs informative negative samples based on the prior knowledge of rare but dangerous circumstances.

Decentralized Structural-RNN for Robot Crowd Navigation with Deep Reinforcement Learning

Decentralized structural-Recurrent Neural Network (DS-RNN), a novel network that reasons about spatial and temporal relationships for robot decision making in crowd navigation, is proposed and demonstrated that it outperforms previous methods in challenging crowd navigation scenarios.

Multimodal Deep Generative Models for Trajectory Prediction: A Conditional Variational Autoencoder Approach

This work provides a self-contained tutorial on a conditional variational autoencoder (CVAE) approach to human behavior prediction which, at its core, can produce a multimodal probability distribution over future human trajectories conditioned on past interactions and candidate robot future actions.

Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning

This work introduces Bootstrap Your Own Latent (BYOL), a new approach to self-supervised image representation learning that performs on par or better than the current state of the art on both transfer and semi- supervised benchmarks.

Trajectron++: Dynamically-Feasible Trajectory Forecasting with Heterogeneous Data

Trajectron++ is a modular, graph-structured recurrent model that forecasts the trajectories of a general number of diverse agents while incorporating agent dynamics and heterogeneous data and outperforming a wide array of state-of-the-art deterministic and generative methods.