• Journal of Internet of Things and Convergence
• /
• v.7 no.4
• /
• pp.29-34
• /
• 2021

In order to apply reinforcement learning to a robot in a real environment, it is necessary to use simulation in a virtual environment because numerous iterative learning is required. In addition, it is difficult to apply a learning algorithm that requires a lot of computation for a robot with low-spec. hardware. In this study, ML-Agent, a reinforcement learning frame provided by Unity, was used as a virtual simulation environment to apply reinforcement learning to the obstacle collision avoidance problem of mobile robots with low-spec hardware. A DQN supported by ML-Agent is adopted as a reinforcement learning algorithm and the results for a real robot show that the number of collisions occurred less then 2 times per minute.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.13 no.6
• /
• pp.1006-1012
• /
• 2010

For the autonomous navigation of an unmanned ground vehicle in the rough terrain and combat, the dust environment should necessarily be overcome. Therefore, we propose a robust obstacle detection methodology using laser range sensor and radar. Laser range sensor has a good angle and distance accuracy, however, it has a weakness in the dust environment. On the other hand, radar has not better the angle and distance accuracy than laser range sensor, it has a robustness in the dust environment. Using these characteristics of laser range sensor and radar, we use laser range sensor as a main sensor for normal times and radar as a assist sensor for the dust environment. For fusion of laser range sensor and radar information, the angle and distance data of the laser range sensor and radar are separately transformed to the angle and distance data of virtual range sensor which is located in the center of the vehicle. Through distance comparison of laser range sensor and radar in the same angle, the distance data of a fused virtual range sensor are changed to the distance data of the laser range sensor, if the distance of laser range sensor and radar are similar. In the other case, the distance data of the fused virtual range sensor are changed to the distance data of the radar. The suggested methodology is verified by real experiment.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.53 no.5
• /
• pp.309-314
• /
• 2004

In the established virtual-reality system, although it is possible to transact a faculty of sensation and graphic in a single PC, virtual object forcibly treated with rigid body for the reason of the huge amount of calculation, and the number of polygon is restricted. Furthermore, there is some difficulty in the financial aspect and a program field, because the existing virtual-reality system needs at least two workstations or super computers. In this study, the new force-reflecting algorithm called as "Proxy" and a finite element method of Hyperion are applied to this system in order to transact in real-time. Consequently, though the number of polygon, which brings about an obstacle is increased in the real-time graphic transaction, this system makes it possible to transact in the real-time, not being influenced by the size of the virtual object.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.65 no.2
• /
• pp.342-348
• /
• 2016

We propose an obstacle detection algorithm for unmanned ground vehicle on uneven terrain. The key ideas of the proposed algorithm are the use of two-layer laser range data to calculate the gradient of a target, which is characterized as either ground or obstacles. The proposed obstacle detection algorithm includes 4-steps: 1) Obtain the distance data for each angle from multiple lidars or a multi-layer scan lidar. 2) Calcualate the gradient for each angle of the uneven terrain. 3) Determine ground or obstacle for each angle on the basis of reference gradient. 4) Generate a new distance data for each angle for a virtual laser scanner. The proposed algorithm is verified by various experiments.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.1621-1624
• /
• 2003

This paper presents a new method that generates a path that has no collision with the obstacles or the characters by using the three motion parameters, and automatically creates natural motions of characters that are confined to the path. Our method consists of three parameters: the joint information parameter, the behavior information parameter, and the environment information parameter. The joint information parameters are extracted from the joint angle data of the character and this information is used when creating a path following motion by finding the relation-function of the parameters on each joint. A user can set the behavior information parameter such as velocity, status, and preference and this information is used for creating different paths, motions, and collision avoidance patterns. A user can create the virtual environment such as road and obstacle, also. The environment is stored as environment information parameters to be used later in generating a path without collision. The path is generated using Hermit-curve and each control point is set at important places.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.2461-2466
• /
• 2003

This paper presents a simple sensor fusion algorithm using neural network for navigation of mobile robots with obstacle avoidance and trap recovery. The multiple sensors input sensor data to the input layer of neural network activating the input nodes. The multiple sensors used include optical encoders, ultrasonic sensors, infrared sensors, a magnetic compass sensor, and GPS sensors. The proposed sensor fusion algorithm is combined with the VFH(Vector Field Histogram) algorithm for obstacle avoidance and AGPM(Adaptive Goal Perturbation Method) which sets adaptive virtual goals to escape trap situations. The experiment results show that the proposed low-level fusion algorithm is effective for real-time navigation of mobile robot.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.3 no.4
• /
• pp.244-250
• /
• 2008

This paper describes the design and implementation of an unmanned ground vehicle (UGV) and also estimates how well autonomous navigation and remote control of UGV can be performed through the optimized arbitration of several sensor data, which are acquired from vision, obstacle detection, positioning system, etc. For the autonomous navigation, lane detection and tracing, global positioning, and obstacle avoidance are necessarily required. In addition, for the remote control, two types of experimental environments are established. One is to use a commercial racing wheel module, and the other is to use a haptic device that is useful for a user application based on virtual reality. Experimental results show that autonomous navigation and remote control of the designed UGV can be achieved with more effectiveness and accuracy using the proper arbitration of sensor data and navigation plan.

• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.6
• /
• pp.513-520
• /
• 2011

This paper presents the motion planning of robotic vehicles for the path tracking and the obstacle avoidance. To follow the given path, the vehicle moves through the turning radius obtained through the pure pursuit method, which is a geometric path tracking method. In this paper, we assume that the vehicle is equipped with a 2D laser scanner, allowing it to avoid obstacles within its sensing range. The turning radius for avoiding the obstacle, which is inversely proportional to the virtual force, is then calculated. Therefore, these two kinds of the turning radius are used to generate the steering angle for the front wheel of the vehicle. And the vehicle reduces the velocity when it meets the obstacle or the large steering angle using the potentials of obstacle points and the steering angle. Thus the motion planning of the vehicle is done by planning the steering angle for the front wheels and the velocity. Finally, the performance of the proposed method is tested through simulation.

• 제어로봇시스템학회:학술대회논문집
• /
• 2002.10a
• /
• pp.38.2-38
• /
• 2002

1. Introduction 2. Potential theory and Local minimum problem 3. Discrete modeling method 4. Virtual obstacle concept and extra potential function 5. Simulations 6. Conclusions

• Journal of the Korean Society for Railway
• /
• v.20 no.2
• /
• pp.173-181
• /
• 2017

The obstacle deflector sweeps obstacles off the track or absorbs crash energy with an energy absorber to prevent derailment of a train and to minimize damage and casualties after an accident. In this study, an obstacle deflector and its operational mechanism were designed with a tension-type energy absorber and a 4-bar linkage system. Also, a test method was suggested and verified with FEA (Finite Element Analysis) and UTM (Universal Test Machine) for testing of the static load and energy absorbing ability according to EN 15227 regulations. Through this study, an obstacle deflector that meets the EN 15227 standard was designed and a test method was suggested to adjust the collapse load easily and to verify it experimentally according to the design and verification procedure of the obstacle deflector.