The Obstacle Avoidance System Based DSP For Deep-seabed Tracked Vehicle

Resource issue is one of the problems faced by the human to solve today, so it is significant that mining the rich resources lying deep-seabed. As an important component of deep-seabed mining system, deep-seabed tracked vehicle is used for collecting the nodules in 6000m-depth seabed. Being working in an unknown environment, to ensure high exploitation efficiency, a stable obstacle avoidance system has to be designed to guarantee tracked vehicle's real-time obstacle avoidance during its operation.In this thesis, a DSP-based obstacle avoidance system for deep-seabed tracked vehicle is researched. A distance measurement system based multi-ultrasonic sensors is designed through using 10 ultrasound sensors to detect the obstacles information. As the uncertain information of ultrasonic sensors, BP network is used to do data fusion, the information of obstacles sensors detected is inputted to the network, then more accurate distance can be provided to the deep-seabed tracked vehicle. A genetic-fuzzy approach is presented for local obstacle avoidance of deep-seabed tracked vehicle. According to the obstacle avoidance system's requirements, a fuzzy controller optimized by the genetic algorithm is designed for the vehicle's real-time local obstacle avoidance. Improved genetic algorithm is used to learn fuzzy control rules offline, and a particular fitness function is used to assess the vehicle's movement, then the real-time obstacle avoidance and the shortest path requirement is achieved, and the ability to learn which the traditional fuzzy control does not have is obtained. At last, a DSP-based obstacle avoidance system is designed. By using high-speed processing power of DSP as the main controller, the system's speed is greatly improved.The simulation of the traditional fuzzy control and the genetic-fuzzy approach are processed in the Delphi environment, the results of simulation experiment indicate the effectiveness of the proposed method, and the shortest path requirement can be fulfilled.