Researches on obstacle avoidance based on environment map of unknown environment are not widely applied for walking robots, especially for large scale robots withhydraulically-driven actuators. In contrast, the walking robots are mainly applied to perform specific tasks in a predefined environment. This research aims to improve the capabilities and increase autonomy of the hydraulically-driven hexapod robot COMET-TV by improving mapping technique for unknown environment, obstacles avoidances and leg motion control assistance using a laser range finder (LRF) 3-D point clouds data. The COMET-IV can be controlled from a remote place by an operator, but the operator has to know the surrounding area of the robot and its current conditions to determine the next walking command while steering a robot from a remote place. Therefore, a map of unknown environment is needed, and it is developed using the Occupancy Grid Map (0GM), but the cells are categorized into not only two categories as current existed achievement but multiple categories. On the other hand, for autonomous operation (the scope of this research), the information associated with the map is used as a reference to generate a walking path for robot. Moreover, in order to capitalize the capabilities of the robot, the Grid-based Walking Trajectory for Legged Robot (GWTLR) method is proposed to avoid, walk over and cross over an obstacle, including ascend and descend a cliff with support of proposed Grid-cell Edge Detection method to analyze the obstacle geometrics concerning the map of an unknown environment. The GWTLR method determines the height of the COMET-TV body, leg swing height and leg stride length, and where the robot should stop before its legs are moved to enable the tasks to be performed without collision with the obstacles. in addition, a proposed Grid-based Walking Assistant for Legged Robot (GWALR) method cascaded to the force control and impedance control as a dynamic input reference to increase the robustness for robot walking on unstructured terrain.Experiment results of the proposed methods show that the trajectory planning can be done autonomously under the unknown environment, and it is also demonstrated to be effective to provide the surrounding environment map to the remote operator. Therefore, the proposed methods were proven to be highly potential to be applied for as a part of the overall system for actual stochastic terrain navigation.
展开▼
