Abstract

Path planning is a significant and challenging task in the domain of automatic driving. Many applications, such as autonomous driving, robotic navigation and aircraft object tracking in complex and changing urban road scenes, need accurate and robust path planning by detecting obstacles in the forward direction. The traditional methods only rely on the path search method without considering the environmental factors, the vehicle path planning method cannot deal with the complex and changeable environment. To deal with above problems, we propose a perception-aware based multi-modal feature fusion approach that combines visual-inertial odometer (VIO) poses and semantic obstacles in the forward scene of vehicles to plan driving paths. The proposed method takes environment awareness as the guide and combines path search algorithm to realize path optimization task in complex environment. The proposed approach first uses a long short memory network (LSTM) to build a VIO that fuses visual and inertial data for pose estimation. To detect obstacles, the proposed method uses a segmentation model with a lightweight structure to extract semantic 3D landmarks. Finally, a path search strategy combining an A* algorithm and visual information is proposed to plan driving paths for intelligent vehicles. We estimate the proposed path planning method on assimilated scenes and public datasets (KITTI and Cityscapes) by using a micro controller (Jetson Xavier NX) installed on a small vehicle. We also show comparable results with path planning that only uses the greedy algorithm or heuristic algorithm without using visual information and show that our method is adequate in coping with different complex scenes.

Highlights

  • Visual perception is an essential part of automobile intelligence

  • We introduce the long short memory network (LSTM) to build a visual-inertial odometer (VIO), which is used for pose estimation

  • The proposed method carries out path planning task based on the scene perception and path searching

Read more

Summary

INTRODUCTION

Visual perception is an essential part of automobile intelligence. As the autonomous driving industry continues to advance, the demands of perception-related technologies are growing continuously, which undoubtedly pushes forward the rapid development of technologies. The existing approaches mainly include VIFTrack [13], Maplab [14], Iceba [15], and PIVO [16], which almost always use fused data for some recognition tasks such as feature tracking, automatic navigation or pose estimation By comparing these approaches with a single sensor approach, we can see that VIOs have better performance for visual recognition tasks. These methods can only cope with problems of localization and navigation and are unable to cope with the problem of path planning for automatic driving. 4) The proposed method is evaluated on a public urban street scene benchmark and a simulative environment based on a Jetson Xavier NX installed on a small vehicle using the standards of existing methods

RELATED WORK
POSE ESTIMATION WITH A FUSION OF MULTI-MODAL NEURAL FEATURES
SEMANTIC 3D OBSTACLE DETECTION
EXPERIMENTAL RESULTS AND ANALYSIS
SEMANTIC SEGMENTATION FOR OBSTACLE DETECTION
PATH PLANNING FOR OBSTACLE AVOIDANCE
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call