Path Planning

Path planning technology is a key branch of robotics research. Optimal path planning involves finding an optimal path from a starting state to a target state within a robot's workspace, avoiding obstacles, based on one or more optimization criteria (such as minimizing work cost, shortest travel distance, and shortest travel time).

Path planning methods can be broadly categorized into two types: traditional and intelligent. Traditional path planning methods primarily include the free-space method, graph search, grid decoupling, and artificial potential field methods. Most global planning in robot path planning is based on these methods, but these methods need further improvement in terms of path search efficiency and path optimization. The artificial potential field method is a relatively mature and efficient planning method among traditional algorithms. It uses an environmental potential field model for path planning, but does not examine whether the path is optimal.

Intelligent path planning approaches apply artificial intelligence methods such as genetic algorithms, fuzzy logic, and neural networks to path planning. This approach improves the accuracy of robot path planning and accelerates planning speed to meet practical application needs. Fuzzy methods, neural networks, genetic algorithms, Q-learning, and hybrid algorithms are among the most widely used algorithms. These methods have achieved significant research results in both known and unknown obstacle environments.

Robot Vision

The vision system is a crucial component of autonomous robots, typically consisting of a camera, an image acquisition card, and a computer. Robotic vision systems perform tasks including image acquisition, image processing and analysis, and output and display. Core tasks include feature extraction, image segmentation, and image recognition. Accurately and efficiently processing visual information is a key issue for vision systems. Visual information processing has become increasingly specialized, encompassing compression and filtering of visual information, environment and obstacle detection, recognition of specific environmental landmarks, and three-dimensional information perception and processing. Environment and obstacle detection is both the most important and challenging process in visual information processing. Edge extraction is a commonly used method in visual information processing. Conventional methods for image edge extraction, such as those based on local data gradients and second-order differentials, are incapable of meeting the real-time requirements of mobile robots that must process images while in motion. Consequently, computational intelligence-based image edge extraction methods have been proposed, such as those based on neural networks and those utilizing fuzzy inference rules. In particular, Professor J.C. Bezdek recently comprehensively discussed the significance of using fuzzy logic for image edge extraction. Specifically for visual navigation, this approach integrates the road knowledge needed for outdoor robot movement, such as white lines and road edges, into a fuzzy rule base to improve road recognition efficiency and robustness. Others have proposed combining genetic algorithms with fuzzy logic.

Robot vision is one of the most important indicators of its intelligence and is of great significance to both robot intelligence and control. Research is underway both domestically and internationally, and several systems have already been put into use.

Intelligent control

With the development of robotics, traditional control theory has exposed its shortcomings for physical objects that cannot be accurately analytically modeled and pathological processes with insufficient information. In recent years, many researchers have proposed various intelligent control systems for robots. Intelligent control methods for robots include fuzzy control, neural network control, and the fusion of intelligent control technologies (including the fusion of fuzzy control and variable structure control, the fusion of neural networks and variable structure control, and the fusion of fuzzy control and neural network control; intelligent fusion technologies also include fuzzy control methods based on genetic algorithms).