Research On Evaluation Method In Automatic Stake Test Based On Vision Of Autonomous Mobile Robot

The vision technology of Autonomous Mobile Robot (AMR) recently becomes a research hotspot in robot field, which has wide application prospect in intelligent human-machine interactive systems,etc. For the sake of equality and accuracy of Automatic Stake Test in Driving License Examination and trying to improve the function of electronic Stake Test System, in this thesis the vision technology is applied to Automatic Stake Test System. It is deeply studied of the AMR examiner system which has imitation of artificial visual function.Partial contents of paper:1. The design of decision-making module of AMR in its imitation of artificial supervision of the tested car. By applying Finite States Machine (FSM) theory to the design, the abstract of the whole process of the Automatic Stake Test can reveal the states and events that a real person could meet during the test, which is to be presented in state diagram. Later they will be minimized with redundancy states deleted. Finally the FSM of AMR its imitation of artificial supervision of the tested car is proposed. So using FSM makes the design more easy and understandable.2. The research regarding route identification of the tested car in Automatic Stake Test. Thanks to the vision sensor, the relative-position information between the tested car and the sideline as well as the angle between axis and the line that connect the particles in two adjacent images is obtained, accordingly a method of route-identification is promoted to help judge whether the tested car break the rule or not.3. The research into detection of tested-car and identification algorithm in the Automatic Stake Test. Inspired by combination of improved gray-projection algorithm and improved HSI difference, a method to detect moving objects with no special background is developed to gain examples and experiment result. Such algorithm is characterized by strong robustness and good real-time. According to volume characteristics and profile characteristics of the moving objects and so on, an algorithm to classify and recognize moving object is provided, which can quickly and accurately identify the moving object and provide the necessary conditions for AMR to track and judge the tested car.4. The design of a system in AMR's automatic tracking of the tested car. Such system designed from an approach combining the image measure and the predictive control theory, tested car can be tracked in real-time. 5. The design of a digital prototype system about AMR examiner in the Automatic Stake Test system. The states and events in the whole process of the Automatic Stake Test are revealed and the evaluation method is promoted. By simulating and enumerating the states and events in the whole process of the Automatic Stake Test, it is proved that the system can achieve the function of monitoring and evaluating the tested car with the AMR examiner imitating the human.