Mobile robots mainly need to solve the problems of positioning, planning, control, etc. At present, the key research fields include environmental awareness and modeling, positioning and navigation, environmental understanding, multi robot coordination, etc. In the future, mobile robots will develop towards the following trends:
"Natural navigation+independent path planning" has become the mainstream
The development of mobile robots has gone through different stages of track based mode (such as tape traction mode), beacon mode (such as QR code), and beacon free mode (such as SLAM, real-time positioning and map building). SLAM technology can enable robots to achieve positioning and navigation without beacons. It is easy to deploy, flexible, and more suitable for applications in complex operating environments and frequently changing business scenarios. Therefore, it is favored by more and more customers and is becoming the mainstream trend in the industry.
Industry development shows that the development of navigation technology makes the equipment gradually transition from "car" to "robot". With the development of new technology, AGV has become more and more autonomous and intelligent, and the evolution of AMR has expanded the application of the industry.
At this stage, there is no one navigation mode that can "conquer the world". The most suitable navigation mode can only be selected according to the characteristics of the application. Different applications have different requirements for navigation. Among all kinds of navigation methods, the most popular ones are laser, vision and other natural navigation methods that do not rely on artificial environment.
The diversity of applications determines the diversification of technology development directions. The standards for measuring the advantages and disadvantages of technology vary according to different application needs. It is difficult to use a unified standard to measure different technologies.
Deep learning will be widely used to enhance the robot's understanding of the surrounding environment
The application of depth learning technology in AI in computer vision mainly includes object recognition, object detection and tracking, semantic segmentation, instance segmentation, etc. Semantic SLAM can combine object recognition with visual SLAM, introduce label information into the optimization process, build maps with object labels, and realize the robot's understanding of the content of the surrounding environment.
Traditional 2D obstacle detection has many limitations. Artificial intelligence semantic segmentation can more effectively judge the situation of people or obstacles, improve the detour efficiency, and robot system can improve the application efficiency and intelligent level.
The accelerated integration of new technology and robot technology will further promote the upgrading of products. The autonomy of mobile robot is mainly embodied in three aspects: "state awareness", "real-time decision-making" and "accurate implementation". The Internet of Things, AI, 5G and other new generation information technologies are combined with robot technology to enable efficient interaction of devices, more free flow of data, and maximize the effectiveness of hardware command through algorithms.