Behavior-based navigation of a two-wheeled self-balancing robot using a modified hybrid automaton

Abstract

Due to advances in robotics science, mobile robots are being used in more and more applications worldwide, and the autonomous navigation of these robots is an important topic in their discussion. This paper focuses on the autonomous navigation of a two-wheeled self-balancing robot (TWSBR) in an unknown environment using behavior-based control in the form of a hybrid automaton. This hybrid automaton includes the behaviors “Go To Goal” and “Avoid Obstacle,” and to avoid the Zeno phenomenon between these two behaviors, another behavior is considered in between, called “Follow Wall,” which the robot uses to move around the obstacle. However, two bugs are identified in the conventional hybrid automaton. The first bug causes the robot to not follow the optimal path. Another bug is that the Zeno phenomenon occurs between the two behaviors “Follow Wall” and “Go To Goal,” causing odometry errors in the experimental environment. The results show that the modified hybrid automaton successfully corrects the bugs and works as intended. The navigation algorithm is designed for the point mass model, so it is transformed to the unicycle model using a transformation, which can be used as input to the TWSBR controller. After linearizing the dynamic equations of the robot around its equilibrium point, the pole placement method is used to create the TWSBR controller. By adding the Luenberger observer to estimate the state variables, the non-full-state feedback system is also controlled. The results of the simulations demonstrate that the whole system is functioning properly so that the robot follows the path determined by the navigation algorithm while maintaining its equilibrium.