Research On Simultaneous Localization And Mapping Methods For Snake-like Robots

Snake-like robots are bionic robot designed to imitate the biological snakes.They have excellent maneuverability and can adapt to narrow environment with its slender shape advantage and flexible gaits brought by redundant joints.It is widely used in disaster rescue,underground exploration and other application scenarios that traditional robots cannot adapt to.Environmental modeling and self-locallization are the primary functions for the snake-like robots to perform tasks autonomously in the above environment.However,the current environmental modeling and positioning methods for the snake-like robots are very limited,and the scheme relying on external positioning system cannot adapt to the actual working conditions of the snake-like robots.Therefore,it is necessary to equip the snake-like robots with specific sensors to achieve environmental modeling without prior environmental information,and dynamically estimate its own position during the motion process,which is simultaneous localization and mapping(SLAM).However,the large turning angles,fast rotation speed,and other motion characteristics of snake-like robots during their movement can lead to problems such as drift or even failure in localization and mapping.Therefore,this thesis proposes two methods to improve the localization and mapping accuracy of the snake-like robots from the aspects of improving the gait and improving the robustness of visual SLAM algorithm.The research work of this thesis is as follows:(1)Research on adaptability of SLAM and typical gaits of snake-like robots.The SLAM adaptability of different types of typical gaits was analyzed,taking into account issues related to large turning angles and fast rotation speed during the robots’ movement process.To achieve this,the Undulating motion,Rolling gait,Pedal gait,and Sidewinding gait were analyzed with regard to their unique movement characteristics,the visual SLAM algorithm based on feature extraction and matching was introduced and a visual SLAM simulation system was established to evaluate the SLAM adaptability of the basic gaits of snake-like robots from three aspects: whether feature points can be extracted,whether there is drift,and the size of absolute error between the estimated trajectory and the true trajectory.(2)From the perspective of improving the inherent characteristics of snake-like robots gaits,a visual SLAM method based on head stability control for snake-like robots was proposed.The Undulating motion was selected as the target for gait improvement,and the principle of head stability control was explained.By comparing the movement states of the head joint under different waveform amplitudes,the effectiveness of head stability control in controlling the stability of the head joint of the snake-like robot under the undulating gait was verified.A simulation analysis of the snake-like robot localization and mapping was also carried out under this control.(3)Research on visual-inertial tightly coupled SLAM method for snake-like robots.From the perspective of improving the accuracy and robustness of the SLAM algorithm itself,a snake-like robot vision inertia tightly coupled SLAM method is proposed by studying the establishment of factor graphs,IMU pre-integration,and initialization theories.The effectiveness of the proposed method in enhancing the robustness of the visual SLAM algorithm and improving the accuracy of snake-like robot localization and mapping was verified using simulation analysis tools.(4)Experimental research on simultaneous localization and mapping of the snakelike robot.The physical experiment platform required by the experiment was built and the internal and external parameters of the camera and IMU were calibrated.Experimental research on simultaneous localization and mapping of snake-like robots under Undulating motion and Arc Rolling gait was carried out,which mainly included:comparison experiment of head joint motion state of snake-like robots under Undulating motion at different wave amplitudes;comparison experiment on the influence of head stability control method on the accuracy of localization and mapping of snake-like robots under comparison experiment;localization and mapping experiment of snake-like robots based on visual-inertial SLAM method.The experimental results show that the head stability control method and the visual-inertia tightly coupled method can effectively improve the localization and mapping accuracy of the snake-like robots.The thesis has 99 figures,3 tables and 80 references.