Perception-driven model for hyper-redundant mobile robot navigation in unstructured environment

Abstract

Navigation in various terrains is considered one of the biggest challenges in the robotics field. Hyper-redundant robots are designed to imitate the feature of biological snakes. Their flexible and multi-degree of freedom body give them an advantage over other traditional mobile robots. Most of the studies that have been done considered wheel snake robots that navigate using horizontal motion in environments with the obstacle. However, irregularity of the surface is challenging for these types of robots. Most of the past works on snake robot locomotion have almost exclusively considered motion across smooth surfaces. However, many real-life environments are not smooth but cluttered with obstacles and irregularities. When the operational scenario is characterized by a surface that is no longer assumed to be flat and obstacles are present, snake robots can move by sensing the surrounding environment. This thesis makes progress towards addressing these issues by developing a perception-driven model for navigation of the hyper robotic system in an unstructured environment. The proposed model should be able to detect the surrounding operational space, identifying the walls, obstacles, and other external objects to create local motion planning. In this thesis, the modeling and design of hyper-redundant mobile robot are presented. The presented model is analyzed using both Newton-Euler and Euler-Lagrange equations. The mathematical model of traveling wave locomotion is verified through simulation and experiment. The simulation shows the relationship between the joint's torque and the position of each joint from the center of gravity. The simulation also describes the effects of the environment, such as the coefficient of friction and initial winding angle on the joint's torque. One of the most challenging of this gait is to stabilize the robot during propulsion. In this research, stability is achieved by controlling the body shape of the robot by controlling the lateral joints of the robot. The main contribution of the thesis is that the development of hyper-redundant mobile robots can navigate in unstructured terrain. A laser range finder sensor (LRF) is used to estimate the distance between the robot and the obstacle. To optimize the sensory signals, undesired detected obstacles points in the scanning zone was eliminated by the established obstacles' filtering. A guidance system is developed to control the navigation of the robot in an unstructured environment. In the system, the robot mode switches between traveling waves, obstacle avoidance, and climbing obstacle locomotion according to the type of the environmental model. The robot system was verified by analyzed the performances in different environments such as narrow path, environment with varying types of obstacles (box, crumple papers, and walls), and steps. Step ascending and descending is achieved by a group of customized designed steps. The perception system could detect and recognize the obstacles’ type and achieve successful obstacle avoidance both in narrow and cluttered terrain. The results also showed that the robot could observe and climb the step of 0.1 m height successfully with an average time of 0.9 min.