Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Intention-based and Risk-Aware Trajectory Prediction for Autonomous Driving in Complex Traffic Scenarios

Wen Wei, Jiankun Wang

TL;DR

The paper tackles trajectory prediction for autonomous driving in complex traffic by introducing a joint prediction framework composed of Interaction, Intention, and Risk Assessment modules. The interaction module models dynamic multi-agent dependencies, the intention module narrows predictions to plausible driving behaviors, and the risk assessment module biases trajectory optimization toward safer outcomes via a risk-ethics-based cost. Empirical results on the DeepAccident dataset show substantial gains over state-of-the-art methods in both normal and accident scenarios, with ablations confirming the importance of each module, especially the risk-aware component. The work advances practical autonomous driving by combining behavior-aware prediction with risk-sensitive planning, and provides code for reproducibility.

Abstract

Accurately predicting the trajectory of surrounding vehicles is a critical challenge for autonomous vehicles. In complex traffic scenarios, there are two significant issues with the current autonomous driving system: the cognitive uncertainty of prediction and the lack of risk awareness, which limit the further development of autonomous driving. To address this challenge, we introduce a novel trajectory prediction model that incorporates insights and principles from driving behavior, ethical decision-making, and risk assessment. Based on joint prediction, our model consists of interaction, intention, and risk assessment modules. The dynamic variation of interaction between vehicles can be comprehensively captured at each timestamp in the interaction module. Based on interaction information, our model considers primary intentions for vehicles to enhance the diversity of trajectory generation. The optimization of predicted trajectories follows the advanced risk-aware decision-making principles. Experimental results are evaluated on the DeepAccident dataset; our approach shows its remarkable prediction performance on normal and accident scenarios and outperforms the state-of-the-art algorithms by at least 28.9\% and 26.5\%, respectively. The proposed model improves the proficiency and adaptability of trajectory prediction in complex traffic scenarios. The code for the proposed model is available at https://sites.google.com/view/ir-prediction.

Intention-based and Risk-Aware Trajectory Prediction for Autonomous Driving in Complex Traffic Scenarios

TL;DR

The paper tackles trajectory prediction for autonomous driving in complex traffic by introducing a joint prediction framework composed of Interaction, Intention, and Risk Assessment modules. The interaction module models dynamic multi-agent dependencies, the intention module narrows predictions to plausible driving behaviors, and the risk assessment module biases trajectory optimization toward safer outcomes via a risk-ethics-based cost. Empirical results on the DeepAccident dataset show substantial gains over state-of-the-art methods in both normal and accident scenarios, with ablations confirming the importance of each module, especially the risk-aware component. The work advances practical autonomous driving by combining behavior-aware prediction with risk-sensitive planning, and provides code for reproducibility.

Abstract

Accurately predicting the trajectory of surrounding vehicles is a critical challenge for autonomous vehicles. In complex traffic scenarios, there are two significant issues with the current autonomous driving system: the cognitive uncertainty of prediction and the lack of risk awareness, which limit the further development of autonomous driving. To address this challenge, we introduce a novel trajectory prediction model that incorporates insights and principles from driving behavior, ethical decision-making, and risk assessment. Based on joint prediction, our model consists of interaction, intention, and risk assessment modules. The dynamic variation of interaction between vehicles can be comprehensively captured at each timestamp in the interaction module. Based on interaction information, our model considers primary intentions for vehicles to enhance the diversity of trajectory generation. The optimization of predicted trajectories follows the advanced risk-aware decision-making principles. Experimental results are evaluated on the DeepAccident dataset; our approach shows its remarkable prediction performance on normal and accident scenarios and outperforms the state-of-the-art algorithms by at least 28.9\% and 26.5\%, respectively. The proposed model improves the proficiency and adaptability of trajectory prediction in complex traffic scenarios. The code for the proposed model is available at https://sites.google.com/view/ir-prediction.
Paper Structure (15 sections, 13 equations, 4 figures, 3 tables)

This paper contains 15 sections, 13 equations, 4 figures, 3 tables.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORK
  3. PRELIMINARIES
  4. PROPOSED MODEL
  5. Interaction Module
  6. Intention Module
  7. Risk Assessment Module
  8. risk estimation
  9. cost function
  10. Model Training
  11. EXPERIMENTS
  12. Experiment Setup
  13. Results
  14. Ablation Study
  15. CONCLUSIONS

Figures (4)

  • Figure 1: i@ represents the ego vehicle, ii@ represents the other vehicle. This figure illustrates the process from trajectory generation to optimization in a multi-vehicle scenario. Our model generates various feasible vehicle trajectories, which are adjusted based on intention prediction. If the model detects that the AV's trajectory intersects with a high-risk area, such as a sidewalk, the trajectory is optimized to ensure both safety and adaptability.
  • Figure 2: Illustration of the proposed model.
  • Figure 3: The proportion of different scenario subsets.
  • Figure 4: Illustration of prediction results in different traffic scenes. Top row: scenario exists for one right-turn operation. Bottom row: scenario exists collision probability. Compared with baselines, our model performs better in both scenarios, not only predicting accurately but also considering potential risks in the future.