Subjective Uncertain Knowledge Representation And Robot Path Planning Algorithm Based On Grey Qualitative Method

Due to the complexity of the objective environment or the cognitive ability limit of the intelligent systems, the knowledge gained from the environment by the intelligent systems is uncertain. The research on how to simulate human intelligence to represent and process the uncertain knowledge and represent it formally, endowing the robot the ability to process uncertain knowledge, is one of the hot spot in AI. During the perception process for environment, human always gather environment knowledge step by step and select the most important information for storing. This fact makes the subjective uncertain knowledge which stems from the incompleteness of knowledge become one of the most important uncertain knowledge in human memory. At present, mobile robot gradually step into the daily life of human beings. So it needs the ability to interact with human, the ability to represent and use the subjective uncertain knowledge which similar to human intelligence. As a result, research on theory and methods of how to represent subjective uncertain knowledge and the corresponding reasoning, decision methods have an important significance for improvement of mobile intelligence.Grey system theory is aimed at represent subjective knowledge. On the other hand, the methods of system modeling and problem analysis method in qualitative theory show the interest in imitating human intelligence. Aiming at establishing a knowledge representation method that complies with human intelligence, and then establishing a method which is both qualitative and quantitative to achieve intelligent reasoning and decision based on the knowledge representation system, the paper proposes a new method-grey qualitative knowledge representation method to represent subjective uncertain knowledge which makes a fusion of grey system theory and qualitative theory. And then, based on the proposed knowledge representation method and the features of human cognitive map, grey qualitative map which combine the superiorities of both cognitive map and navigation map is proposed. The path planning algorithm based on grey qualitative map is also proposed. The experiments results show the advantages in both knowledge complexity and path planning result.The main contributions are as follows:(1) We developed the grey qualitative representation method for representing subjective uncertain knowledge by integrating the features of the grey system theory and the qualitative theory. The method is composed of grey qualitative fundamental element, the set of key points of grey qualitative fundamental element, grey qualitative fundamental element space, grey qualitative relationship, grey qualitative characteristic values and generalized whitening function. Grey qualitative element, grey qualitative fundamental element space and grey qualitative relationship are respectively correspond to ontology primitive, quantity space and causality which are all basic elements in qualitative theory. The grey qualitative fundamental element, the set of key points of grey qualitative fundamental element and the generalized whitening function are correspond to interval grey number, the boundary points of interval grey number and whitening function in grey system theory respectively. We use grey qualitative element as the bridge for the integration of grey system theory and qualitative theory.(2) Show the difference between objective uncertain system and subjective uncertain system. Grey qualitative modeling method is proposed for modeling systems with a small amount of known subjective uncertain rules. The proposed method simulates human intelligence in gathering, fusion and using subjective uncertain knowledge in unknown environment.(3) By investigating the existing research on the cognitive map at home and abroad and the grey qualitative knowledge representation method, a grey qualitative map is proposed which suitable for both mobile robot navigation and inaction with human. We define the environment subdivisions and adjacent relationship between then as a qualitative layer of the map, which is used for simulating cognitive map of human intelligence. The quantitative layer including coordinates of vertices of subdivisions and the vector of potential field are used for deciding the robot speed and direction in the navigation process. Another superiority of grey qualitative map is that it can support robot complete path planning task based on a little of key information of environment, which reduce the complexity of environment model.(4) An artificial potential field without traps algorithm based on grey qualitative map is proposed. By calculating the potential field with the key message in grey qualitative map, we solve the local trap problem exists in the traditional artificial potential field which calculated only by repulsive force of obstales and attractive force of target. And further, by adjusting part of vertex coordinates of grey qualitative elements and the potential field vector, the algorithm is optimized. By this, the robot can obtain a smooth path in artificial potential field, which facilitating the practical application.