Range Sensor-Based Efficient Obstacle Avoidance through Selective Decision-Making

Abstract

In this paper, we address a collision avoidance method for mobile robots. Many conventional obstacle avoidance methods have been focused solely on avoiding obstacles. However, this can cause instability when passing through a narrow passage, and can also generate zig-zag motions. We define two strategies for obstacle avoidance, known as Entry mode and Bypass mode. Entry mode is a pattern for passing through the gap between obstacles, while Bypass mode is a pattern for making a detour around obstacles safely. With these two modes, we propose an efficient obstacle avoidance method based on the Expanded Guide Circle (EGC) method with selective decision-making. The simulation and experiment results show the validity of the proposed method.

Extending the Application of an Assistant Personal Robot as a Walk-Helper Tool

“…Robotics 2019, 8, 27 10 of 14 the wheels, and the information of the servomotors was logged in order to enable the representation of the walking trajectory. The experiments have been developed in a simulated household domain and consists in using the walk-helper while conducting simple ordinary tasks such as preparing a simple meal in the kitchen, and going to the living room.…”

“…Robotics 2019, 8, 27 11 of 14 the LIDAR to perform small trajectory corrections to automatically guide the walk while keeping the distance with surrounding obstacles so only one trajectory correction was required to reach the destination and complete the displacement. Comparatively, Figure 6 shows a trajectory composed by a sequence of straight displacements and rotations while Figure 7 show an almost continuous curved trajectory that slightly avoids the obstacles in the path of the displacement.…”

“…Robotics 2019, 8, 27 11 of 14 the LIDAR to perform small trajectory corrections to automatically guide the walk while keeping the distance with surrounding obstacles so only one trajectory correction was required to reach the destination and complete the displacement. Figure 6.…”

“…The general idea of extending the application of the APR as a walk-helper is as follows: the user initiates the walk-helping operation by holding the arms on the mobile robot while the guidance of the walking trajectory is performed by interacting with the arms. The implementation of this general idea is based on the following steps: (1) the user must be located behind the mobile robot; (2) the interaction with one arm of the mobile robot can be used to rotate the APR according to the direction of the destination or to guarantee that the person is located behind the mobile robot; (3) the user must hold both arms and pull them back until reaching a comfortable holding position; (4) after three seconds the mobile robots detects the pulling of both arms, applies power to the servomotors in order to keep the arms in a fixed orientation (the arms are used as a holding support during the walking), and enters into the walk-helper operation; (5) the control of the walking trajectory is then performed by interacting with the arms of the mobile robot (changing the torque applied to each arm); (6) the onboard mobile robot sensors are used to detect and avoid obstacles [27] in the walking trajectory; (7) the onboard mobile robot sensors can be used to supervise the direction of the walking [28] in order to perform imperceptible trajectory corrections without user interaction, for example centering the path of the mobile robot when going through a door; (8) the operation as a walk-helper ends when no holding support or torque is detected in the arms.…”

Learn to Steer through Deep Reinforcement Learning

“…In most of the missions, it is essential for autonomous robots to detect and avoid various obstacles while maneuver in unknown cluttered environments safely. Numerous methods have been proposed and adapted successfully to different robots [ 2 , 3 , 4 , 5 ]. However, conventional methods may impose intensive computational demand [ 6 , 7 ] and are often built upon a set of assumptions that are likely not to be satisfied in practice [ 8 , 9 ].…”