Research And Evaluation On Comprehensive Obstacle-avoiding Behavior For Unmanned Vehicles

Now, technology and information affect every aspect of our daily life, so doautomotive industry. More and more automobile manufacturers all around the world focuson developing unmanned vehicle technology with the help of research institutes andsupported policy from governments, especially in the recent20years. Then testing andevaluation methods on the intelligent behaviors of unmanned vehicles are in urgent need, asthey can regulate unmanned vehicles’ intelligent behaviors and lead to an improvement inthe key technology of unmanned vehicles and advance driver assistant system.This reaearch is supported by the key project of National Natural Science Foundationof China Research on Comprehensive Test Environment Design and EvaluationArchitecture of Intelligent Behaviors of Unmanned Vechiles (project number:90920304)and funding project Design and Implementation of the Verifaction Environment forCognitive Ability Test of Unmanned Vehicles (project number:91120010). With thepurpose of providing proper testing and evaluation methods on comprehensiveobstacle-avoiding behaviors of unmanned vehicles, this research focus on overall analysisof manned vehicles driving behaviors, especially several typical situations, like lane changeconsidering the four surrounding vehicles. Then lane change model was built byconsidering safe distance and speed error between the host vehicle and the surroundingones, front wheel steering angle. In order to validate and optimize the model, virtual visualexperiment platform was set up and testing experiments were operated carefully.Considering driving behaviors’ characteristics, fuzzy comprehensive evaluation andanalytic hierarchy process were taken as the main methods to evaluate obstacle-avoidingbehaviors of unmanned vehicles. Fuzzy rules and evaluation models were established byanalyzing virtual visual experiments data, then unmanned vehicles obstacle-avoiding datafrom recent Future Challenge were taken into our evaluation system. The output wascompared with the game results and the validity of this reaearch was verified. This researchwill be a reference for next Future Challenge and also lead to an improvement inunmanned vehicle technology and advance driver assistant system.