Research Status and Development Trend of Intelligent Wheelchair Navigation System

0 Preface

As a service robot in the field of medical care, the intelligent wheelchair uses a lot of mobile robot technology. The key technologies involved in the research of intelligent wheelchairs are navigation system, control and energy system, and human-machine interface . However, since the entire wheelchair system is human-centered, the core of the research is to solve the problem of safe navigation of wheelchairs. The so-called navigation means that the mobile robot makes global path planning according to the known map information according to the predetermined task command, and continuously perceives the surrounding local environment information during the traveling process, and makes various decisions autonomously and adjusts at any time. In your own position, guide yourself to safe driving to the target position.

This article refers to the address: http://

This paper analyzes the core issues in intelligent wheelchair navigation , points out the technical progress and shortcomings involved in the current research, and briefly describes its development trend.

1 system positioning

Intelligent wheelchair positioning is also the acquisition of environmental information. It refers to the use of its own sensor during the movement to determine its position and attitude relative to the global coordinates in the reference coordinate system in the working environment.

Positioning technology can be divided into two categories: machine vision based positioning technology and non-computer sensor based positioning technology. Common positioning methods include optical code disk, inertial gyro, magnetic compass, and road sign matching. Each method has its own advantages and limitations. In practical applications, the actual integration of intelligent wheelchairs uses several methods to improve the accuracy and reliability of the positioning system, but the accuracy is a certain distance from the normal use of the user, so how to improve the positioning accuracy And performance is one of the issues to be studied in the future.

The choice of sensor is important in positioning. Sensors can be divided into visual and non-visual sensors depending on the positioning technology. Currently commonly used sensors include ultrasonic ranging sensors , CCD cameras, infrared sensors , laser sensors , and GPS. Due to the convenience of ultrasonic obstacle avoidance, mature technology and low cost, it has become a common positioning method for intelligent wheelchairs . Multiple ultrasonic ranging sensors are used in the application, and the distance of obstacles is detected by ultrasonic ranging sensors , and then the current position of the robot is determined.

2 Information Fusion Technology

The information obtained from the sensor cannot be guaranteed to be completely reliable and correct, and may cause misjudgment of the physical object or error in the detection of the distance. In this case, we can use the probabilistic method, the integrated multiple observation method, the multi- sensor information fusion method, etc. Processing, in which the research of multi- sensor information fusion method has become a hot topic in recent years.

The so-called information fusion can be broadly summarized as a process of analyzing and combining data and information from multiple sensors into a comprehensive intelligence report according to established rules, and providing demand information to system users on this basis. Such as: decision, mission, track and so on. In sensor information fusion, it is necessary to use a variety of types of sensors . Multi- sensor information fusion technology has demonstrated the superiority of a single sensor , and by synthesis, more reliable information can be obtained than from any single input data.

How to integrate these complementary or redundant sensor information and obtain a more comprehensive information method that reflects the environmental characteristics is particularly important. The most critical part of the research is the research of information fusion algorithms. Many multi- sensor information fusion algorithms applied to different systems have been proposed. These algorithms can be divided into two categories: random class methods and artificial intelligence methods.

(1) Random class method

Such methods are randomized study, random class in a multi-sensor data fusion often includes many methods, such as: the weighted average method, a statistical decision theory, cluster analysis, wavelet transform, Bayes inference, Dempster-Sharer Evidence theory, Kalman filter fusion algorithm, etc.

(2) Artificial intelligence method

In recent years, computational intelligence methods for multi- sensor data fusion include fuzzy set theory, expert systems, neural networks, rough set theory, and support vector machines. Among them, the fusion of various sensor information based on neural networks is a hot spot in recent years. The neural network has good fault tolerance, hierarchy, plasticity, adaptability, associative memory and parallel processing ability. Combining neural network with other methods to carry out information fusion technology research, the effect is remarkable, and a research trend has been formed. Such as wavelet and neural network, Kalman filtering and neural network, Dempster-Shafer evidence theory and neural network, fuzzy clustering and neural network, genetic algorithm and neural network.

Future multi- sensor information fusion technologies mainly focus on the improvement of algorithms and the emergence of new algorithms, the development of micro- sensors and multi-level information fusion.

3 path planning problem

Path planning refers to finding a non-collision path for the intelligent wheelchair from the starting point to the end point in the obstacle environment, and optimizing according to certain principles to find an optimal path. The path planning problem includes two aspects: the first is the establishment of the environmental model; the second is the design of the path planning algorithm.

(1) Establishment of an environmental model

Environmental modeling is the premise of path planning. For statically known environments, there have been many successful research results, and its modeling technology is relatively mature. The problem of path planning in some known or completely unknown environments has not been solved satisfactorily. The root cause is the contradiction between the resolution of the environment and the storage of environmental information.

There are roughly three categories of environmental modeling: network port graph models, grid models, and hierarchical models. The network stew model includes free space method, vertex image method, generalized cone method, etc. It is a high-level description of the environment, which has a large amount of calculation and high requirements on sensor accuracy. The grid model divides the space into grids of the same size. The model is simple to build, but the search space is very large. The hierarchical model models the environment according to the consistency criterion of the data region and the principle of spatial deregulation. This model compresses the search space and is easy to use the sensor. The information is updated on the model.

(2) Path planning algorithm

According to the different understanding of environmental information by intelligent wheelchairs , path planning can be divided into two types: global path planning and partial path planning. Among them, the global path planning needs to know all the messages about the environment, and generate a series of key points as sub-target points to the local path planning system. The local path planning only needs obstacle information close to the robot . During the motion process, the internal environmental information is continuously updated according to the information of the sensor , and a starting point or a sub-target point is determined to the next sub-goal. The preferred path for the point. Compare some path search algorithms and seek better solutions. Further study the situation analysis of activity obstacles and give obstacle avoidance strategies. These two aspects are the main problems to be solved by robot path planning. According to the completeness of the understanding of environmental information, path planning can adopt different algorithms. The algorithms commonly used for global path planning are: viewable method, free space method, and grid method. The methods commonly used in local path planning include artificial potential field method, genetic algorithm and fuzzy logic algorithm.

In recent years, based on these traditional methods, these methods have been further integrated and expanded, such as: genetic algorithm based path planning - two-dimensional path coding problem simplified to one-dimensional path coding problem, fuzzy neural network obstacle avoidance method - - Based on the actual error function and membership function method, the lidar-based path planning method--angle potential field method, virtual force field method--dynamic grid method combined with potential field method.

4 Conclusion

The above is a comprehensive discussion of the methods and ideas involved in the study of intelligent wheelchair navigation , and these research methods are also applicable to mobile robots . The research of intelligent wheelchair with better planning ability, real-time and practicality is the core technology of future research, and also the key and difficult problem in the field of robot research.

It has long-strip shape, and it is mainly used for decorative lighting of building or outline of huge buildings. With a large number of models, it has many advantages such as bright and luminous efficiency, energy saving, rich in color, bright and beautiful patterns, high light efficiency, long working life as long as 50,000 hours, resistant to high temperature and low temperature, non-pollution, etc.. The Wall Washer Light includes seven-color wall washer light, full-color wall washer light and multi-color wall washer light. Its power capability determines the distance of its projection, and it is mainly divided into different levels of 18W, 24W, 36W and 48W, etc.. The power can be changed flexibly according to the requirements. It can be controlled by microchip to achieve many dynamic effects such as gradual change, hopping change, color flashing, random flashing and alternating gradual change.It is widely used in commercial properties, hotels, exterior walls of buildings, squares, staircase lightposts, parks, roads, garden trees, ancient buildings and so on. In recent years, LED Wall Washer Light is commonly used in lighting project.

Led Wall Washer Light

Led Wall Washer Light,Led Outdoor Lighting,Led Wall Washer,Wall Washer Light

Shenyang Haimeng Lighting Co. , http://www.hmwled.com