Abstract
This paper presents a redundant multi-object detection method for autonomous driving, exploiting a combination of Light Detection and Ranging (LiDAR) and stereocamera sensors to detect different obstacles. These sensors are used for distinct perception pipelines considering a custom hardware/software architecture deployed on a self-driving electric racing vehicle. Consequently, the creation of a local map with respect to the vehicle position enables development of further local trajectory planning algorithms. The LiDAR-based algorithm exploits segmentation of point clouds for the ground filtering and obstacle detection. The stereocamerabased perception pipeline is based on a Single Shot Detector using a deep learning neural network. The presented algorithm is experimentally validated on the instrumented vehicle during different driving maneuvers.
Highlights
Self-driving vehicles are experiencing a steadily increasing interest all over the world thanks to the most recent technological development, as witnessed in [1] and [2]
This paper proposes a combined sensor architecture with both stereocamera and Light Detection and Ranging (LiDAR) sensors to enhance the perception pipeline robustness in terms of redundancy of the system
Avoiding the sensor fusion at a sensor level can save computational costs, as no additional algorithms are deployed on the devoted control unit
Summary
Self-driving vehicles are experiencing a steadily increasing interest all over the world thanks to the most recent technological development, as witnessed in [1] and [2]. About 94 % of the road accidents are caused by human errors, according to a recent survey [10] These efforts have been motivated by the promise of preventing accidents, and of reducing emissions and reducing drivingrelated stress [11]. A consistent burden to adoption of driverless vehicles is the lack of public trust, since significant concerns, including but not limited to privacy and cybersecurity, have arisen [4]. Considering this framework, environment perception is a fundamental task for autonomous vehicles, which provides the vehicle a crucial assessment about the driving scenario, including an accurate information about the surrounding obstacles positions [12]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have