![]() With the help of semantic information, the data association is upgraded from the traditional pixel level to the object level. To achieve such goals, robots need to recognize information about objects in the scene, find out their locations and build semantic maps. The traditional VSLAM algorithm can meet the basic positioning and navigation requirements of the robot, but cannot complete higher-level tasks such as “help me close the bedroom door”, “go to the kitchen and get me an apple”, etc. People’s demand for intelligent mobile robots is increasing day by day, which put forward a high need for autonomous ability and the human–computer interaction ability of robots. Interested readers can refer to and other sources in the literature. SLAM based on all kinds of laser radar is not within the scope of discussion in this paper. Therefore, this paper focuses on VSLAM and combs out the algorithms derived from it. VSLAM has the advantage of richer environmental information and is considered to be able to give mobile robots stronger perceptual ability and be applied in some specific scenarios. Due to the advantages of cheap, easy installation, abundant environmental information, and easy fusion with other sensors, many vision-based SLAM algorithms have emerged. In recent years, camera-based VSLAM research has attracted extensive attention from researchers. Similar to human eyes, VSLAM mainly uses images as the information source of environmental perception, which is more consistent with human understanding and has more information than laser SLAM. Compared with VSLAM, because of an early start, laser SLAM studies abroad are relatively mature and have been considered the preferred solution for mobile robots for a long time in the past. Some scholars divide SLAM into Laser SLAM and Visual SLAM (VSLAM) according to the different sensors adopted. SLAM is considered to be the key to promoting the real autonomy of mobile robots. Numerous researchers are committed to making robots more intelligent. With the development of computer technology (hardware) and artificial intelligence (software), robot research has received more and more attention and investment. SLAM, as a basic technology, has been applied to mobile robot localization and navigation in the early stage. ![]() It is mainly used to solve the problem of robot localization and map construction when moving in an unknown environment. SLAM refers to self-positioning based on location and map, and building incremental maps based on self-positioning. Since it was first proposed in 1986, SLAM has attracted extensive attention from many researchers and developed rapidly in robotics, virtual reality, and other fields. Therefore, SLAM (Simultaneous Localization and Mapping), which enables localization and mapping in unfamiliar environments, has become a necessary capacity for autonomous mobile robots. People need the mobile robot to perform some tasks by themselves, which needs the robot to be able to adapt to an unfamiliar environment. Introducing deep learning into the VSLAM system to provide semantic information can help robots better perceive the surrounding environment and provide people with higher-level help. We believe that the development of the future intelligent era cannot be without the help of semantic technology. Later, we focus on the help of target detection and semantic segmentation for VSLAM semantic information introduction. Starting with typical neural networks CNN and RNN, we summarize the improvement of neural networks for the VSLAM system in detail. In addition, we focus on the development of semantic VSLAM based on deep learning. For traditional VSLAM, we summarize the advantages and disadvantages of indirect and direct methods in detail and give some classical VSLAM open-source algorithms. This paper introduces the development of VSLAM technology from two aspects: traditional VSLAM and semantic VSLAM combined with deep learning. Semantic information, as high-level environmental information, can enable robots to better understand the surrounding environment. Deep learning has promoted the development of computer vision, and the combination of deep learning and SLAM has attracted more and more attention. Traditional visionbased SLAM research has made many achievements, but it may fail to achieve wished results in challenging environments. ![]() ![]() Visual SLAM (VSLAM) has been developing rapidly due to its advantages of low-cost sensors, the easy fusion of other sensors, and richer environmental information.
0 Comments
Leave a Reply. |