Tham khảo tài liệu 'mobile robots - moving intelligence part 8', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 14 Biologically-Plausible Reactive Control of Mobile Robots Zapata René Lépinay Pascal Universite Montpellier II France 1. Introduction In many robotic tasks a mobile robot has to perform four subtasks constituting a hierarchy from high level to low level actions Path planning Localization Target pursuit Collision avoidance The last 3 tasks which obviously need perception capabilities and vision like for natural beings can provide the robot with very rich information. Unfortunately artificial vision algorithms present the major disadvantage of being difficult to process in real time for instance the collision avoidance process typical time is less than . Therefore including real time vision in a robot control loop system requires hardware implementation. But efficient reactive behaviours can also be performed with simpler sensors like range sensors Ultrasonic sensors for instance . Biologically-plausible algorithms spring from the observed natural systems especially through analogies between artificial perception actions systems and natural ones Berthoz and Weiss 2000 . Many papers deal with this subject and whole conferences are devoted to it see for instance SAB proceedings SAB 2003 . Neural network theories are an important part of this research and much work has been done on neural aspects of artificial vision. For instance two recent works are closely correlated with biological experiments Wurtz and Lourens 2000 VanRullen 2002 . This chapter is divided into 4 parts. The first section describes a reactive control algorithm based on the modelling of the robot environment interaction. Like Potential fields methods Kathib 1985 or Holenstein and Badreddin 1991 this approach can present the problem of local minima. However it turns out hat it is quite robust to unknown and dynamic worlds. The second section section 3 describes two vision-based algorithms 3D reconstruction and image segmentation which will be the basis of the implementations proposed in the .