Autonomous vehicles rely on perception systems to obtain information about their surroundings. It must detect the presence of other vehicles, pedestrians and other related objects. Due to safety concerns and the need for accurate estimation, lidar systems have been introduced to complement camera- or radar-based recognition systems. This article provides a review of state-of-the-art automotive lidar technologies and the recognition algorithms used in these technologies.
Lidar systems are first introduced by analyzing the key components of these systems, from laser transmitters to beam scanning mechanisms. Introduces and compares the advantages/disadvantages and current status of various solutions. Then, the specific recognition pipeline for processing lidar data is detailed from an autonomous vehicle perspective. Review model-based approaches and new deep learning (DL) solutions. Finally, it provides an overview of the limitations, challenges, and trends of automotive lidar and cognitive systems.
Lidar imaging systems are one of the hottest topics in the optical industry. The need to sense the surroundings of any autonomous vehicle has accelerated the race to determine the final solution to implement. However, the diversity of modern approaches to solutions creates great uncertainty in the determination of the dominant final solution. Moreover, performance data for each approach often comes from different manufacturers and developers with some interest in the dispute. In this paper, we try to overcome the situation by providing a neutral introduction to the technology and its development status related to the lidar imaging system for autonomous vehicles.
It starts with the main single point measurement principle used and is combined with the various imaging strategies described in the article. In fact, an overview of the functions of light sources and photodetectors specific to the most frequently used lidar imaging systems is also provided. Finally, a brief section on pending issues for lidar development in autonomous vehicles has been included to suggest that some issues that still need to be addressed before implementation can be considered final. Readers are provided with detailed bibliography, including both relevant books and recent papers, to further advance the subject.