Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MapsTP: HD Map Images Based Multimodal Trajectory Prediction for Automated Vehicles

Sushil Sharma, Arindam Das, Ganesh Sistu, Mark Halton, Ciarán Eising

TL;DR

This method integrates HD map data to improve the robustness and reliability of trajectory predictions for autonomous vehicles and leverage ResNet-50 to extract image features from high-definition map data and use IMU sensor data to calculate speed, acceleration, and yaw rate.

Abstract

Predicting ego vehicle trajectories remains a critical challenge, especially in urban and dense areas due to the unpredictable behaviours of other vehicles and pedestrians. Multimodal trajectory prediction enhances decision-making by considering multiple possible future trajectories based on diverse sources of environmental data. In this approach, we leverage ResNet-50 to extract image features from high-definition map data and use IMU sensor data to calculate speed, acceleration, and yaw rate. A temporal probabilistic network is employed to compute potential trajectories, selecting the most accurate and highly probable trajectory paths. This method integrates HD map data to improve the robustness and reliability of trajectory predictions for autonomous vehicles.

MapsTP: HD Map Images Based Multimodal Trajectory Prediction for Automated Vehicles

TL;DR

This method integrates HD map data to improve the robustness and reliability of trajectory predictions for autonomous vehicles and leverage ResNet-50 to extract image features from high-definition map data and use IMU sensor data to calculate speed, acceleration, and yaw rate.

Abstract

Predicting ego vehicle trajectories remains a critical challenge, especially in urban and dense areas due to the unpredictable behaviours of other vehicles and pedestrians. Multimodal trajectory prediction enhances decision-making by considering multiple possible future trajectories based on diverse sources of environmental data. In this approach, we leverage ResNet-50 to extract image features from high-definition map data and use IMU sensor data to calculate speed, acceleration, and yaw rate. A temporal probabilistic network is employed to compute potential trajectories, selecting the most accurate and highly probable trajectory paths. This method integrates HD map data to improve the robustness and reliability of trajectory predictions for autonomous vehicles.
Paper Structure (18 sections, 4 equations, 3 figures, 1 table)

This paper contains 18 sections, 4 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Research Background
  3. Our Approach
  4. HD Maps
  5. Feature Extractor
  6. Spatio-Temporal Probabilistic Network (STPN)
  7. Maximum Probability and Output Trajectory
  8. Implementation Details
  9. Experimental setup and Evaluations
  10. Nuscenes Dataset
  11. Evaluation Metrics
  12. MinADE${_k}$
  13. MinFDE${_k}$
  14. MissRate$_{k,d}$
  15. Results
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Standard Autonomous Driving Pipeline Overview
  • Figure 2: Our proposed architecture: HD Map Integration for Vehicle Trajectory Planning. The process begins with the feature extractor analyzing these inputs to create image features. These features are then analyzed by a probabilistic network, which evaluates multiple potential trajectories and assigns probabilities to each.
  • Figure 3: Qualitative Analysis: HD Map Integration for Vehicle Trajectory Prediction. The process begins with the feature extractor analyzing these inputs to create image features. These features are then analyzed by a probabilistic network, which evaluates multiple potential trajectories and assigns probabilities to each. On the left, the HD image is displayed; the middle section shows the multimodal trajectory prediction in red; and on the right, the ground truth trajectory is shown in green.