Strategy And Experimental Research On Robot Odor Source Localization Based On Olfaction Information

As greater and greater importance is attached to the public utilities such as national security and public order, the research on the system of the robot that can replace human beings in the toxic, explosive and other dangerous environments to perform a mission so as to free them from working in danger is of great theoretical and practical significance. At present , in order to make robot do the dangerous work for human beings, the key problem to be urgently solved is how to realize the mobile robot's accurate judgments of the suspicious targets and its quick determination of the source of toxic gas or the location of the suspicious targets.This paper is supported by the Hi-tech Research and Development Program of China -- the Multi-Sensory Robot System of Detecting Dangerous Goods in Extreme Environments (No. 2006AA04Z221). It combines the mobile robot with the multi-sensory humaniod robot head system, and according to the application of seeking the toxic gas source, it makes a deep research into the strategy of robot odor source localization, which lays a foundation for the application of mobile robot to carry out various odor-related dangerous tasks.This paper covers the following original points:1. This paper develops a humaniod robot head system with multi-sensory functions of vision, olfaction and hearing. Based on the analysis of human heads and a comprehensive consideration of such factors as the geometric size of the robot head, the kinematics and dynamics, this paper devises a 4-DOF structure of the humaniod robot head. This structure, with deformed universal joint connected in series, enables the robot head to do three kinds of movement. The scope of flexion and extension movement can reach±30o.The scope of rotary motion can reach±90o,and the scope of side flexion movement can reach±20o. On the basis of CAN bus, a distributed control system with simple structure, good real-time property and high reliability is designed.2. On the basis of bionics, a new kind of robot bionic olfactory system is put forward. This system has a nasal structure similar to that of human beings and an inspiratory system which functions as the human lungs do. It can simultaneously detect CO, SO2, H2S, O2 and one kind of volatile gas of organic compound. The adoption of electrochemistry and photoion gas sensors enables the system to recover quickly and choose correctly, so that the olfactory function of the robot is enormously improved. Moreover, it can conveniently adjust its gas sensor according to the target gas. The system adopts C8051F020 single chip microcomputer to realize the communication with the robot principal controller by means of serial port. With a compact structure, the performance of the system is unaffected by the installation position, so it can be assembled to other type of robots as a robot standardized component , in order to help the robot to complete the olfaction-related work.3. Inspired by the foraging strategies of animals, a strategy of odor source localization is put forward based on the animal predatory behavior. The strategy fuses the information of gas sensor, wind direction sensor and ultrasonic sensor etc. and integrates several good search algorithms such as Z-shaped search, upwind search and fuzzy neural network search and so on. In the strategy, the behavior-based containment control structure is adopted and a behavior module for various search algorithms is designed to enable the robot to select a better search strategy automatically in the course of seeking the odor source. The experimental results show that the strategy has strong robustness, and can meet the requirements of locating odor source in natural indoor environment.4. Drawing on the experience of human beings who also make use of other sense organs apart from the olfaction, this paper proposes a strategy to locate odor source by fusing the olfactory and visual information, and builds an odor source model base that is based on the information fusion of odor, color and distance, to facilitate robot to search and confirm the odor source. For this strategy, this paper designs a control system of the robot olfaction localization on the basis of multi-blackboard model. By seeking and confirming the odor source blackboard, the control system can coordinate the work to realize the odor source localization. Through experiments, this strategy is proved to posses the best efficiency, highest success rate and good ability to adjust to the environment.