Obstacle detection and detour system using line laser and camera

This paper develops a system that can easily and widely detect and avoid obstacles with one system by using a line laser and a camera. In order to use the automatic guided vehicle safely indoors, it is necessary to detect obstacles and avoid collisions. Therefore, we installed a line laser and a camera on the mobile robot to verify the accuracy of the system that detects and avoids obstacles. Conventional automatic guided vehicles include a method of detecting using a plurality of sensors and a method of detecting using LiDAR. However, these have problems that they are easily affected by disturbance and that the sensor itself is expensive and impractical. In this study, by using a line laser and a camera, it is possible to detect easily and widely with one system.


Introduction
In recent years, automated guided vehicles have been used in many fields, including production lines for automobiles and food, and for transporting equipment in hospitals. The global market, including companies that provide automated guided vehicles and related technologies, is expected to grow to $ 4.3 billion by 2027. [1] The current automatic guided vehicle will stop if there is an obstacle on the track. In addition, there is a risk of collision with obstacles. Therefore, it is necessary to detect obstacles and avoid them appropriately. Conventional obstacle detection systems include ultrasonic sensors, PSD sensors, cameras, and LiDAR. However, these have some drawbacks. Ultrasonic sensors and PSD sensors can only measure in one dimension. [2][3] A large number is required for wide-range measurement. In addition, an error occurs due to the influence of ambient noise and reflected light. The method using a camera detects obstacles from the difference between the preset background image and the input image. [4] [5] However, this method requires updating the background image and takes a long time to process. LiDAR is widely used in the field of autonomous driving [6]. However, this has the disadvantages that the sensor itself is expensive and difficult to introduce, and that measurement cannot be performed if the laser is not reflected.
Therefore, the purpose of this research is to develop a system that easily and clearly detects and avoids obstacles using a line laser and a camera.

Obstacle detection
The system configuration diagram is shown in Fig. 1. As shown in Fig. 1, the floor is irradiated with a green line laser and the state is input by the camera. As shown in Fig. 2 front view, when there is an obstacle in front of the robot, the line laser is not straight but interrupted.  Only the line laser is extracted from the image input from the camera. Fig. 3 shows an image of only the line laser extracted.

Obstacle detour
After detecting obstacles, driving efficiency can be improved by avoiding them and running without stopping. If there is an obstacle at the position shown in Fig. 5, the mobile robot determines that the optimum avoidance direction is "left". Let be the line segment connecting the edge of the obstacle and the point at the distance on the extension of the width of the robot.
Step1 The Obstacle detour angle θ [°] is as shown in Eq. (3). The angle θ can be obtained from the right triangle of the distances , .
Step2 The distance can be calculated by the threesquare theorem. The distance [cm] is calculated by Eq.   Fig. 7. The experimental results are shown in Fig. 8. From Fig. 8, the obstacle could be detected from the distance = 79 [cm] between the obstacle and the mobile robot. This factor is based on the principle of detecting obstacles. Since obstacles are detected by the deviation of the line laser, obstacles cannot be detected when the irradiation distance of the line laser is 0 = .  Fig. 8, it was possible to measure accurately without a large error. The maximum error was 1.12 [cm], the minimum error was 0.15 [cm], and the average error was 0.69 [cm]. The average error was less than 1 cm, and it is considered that the detection was accurate. It is considered to be a practical obstacle detection system that can measure in real time.

Obstacle detour
The obstacle avoidance system proposed in this study finds the obstacle avoidance angle θ by using the obstacle width obtained in 3.1 and the distance between the robot and the obstacle. The distance 0 between the line laser irradiation destination and the robot was set to 80 [cm]. The mobile robot moves at a speed of 14.3 [cm / s]. After finding the obstacle avoidable angle θ , the mobile robot changes its angle direction and advances the distance to detour the obstacle. Fig.8 shows how to detour obstacles. Step.1 Step.2 Step.3 Step.4 Step.5 Step.6 Fig. 8 Obstacle detour From Fig. 8, the mobile robot can move while detecting obstacles and avoiding them. After starting the mobile robot, it went straight along the line and detected an obstacle (Step.1). If you continue to go straight on the line, you will collide, so you changed your course to the left, where the movable area is wide (Step.2). At this time, since the width of the obstacle and the distance between the obstacle and the mobile robot are known, the avoidable angle θ was obtained and the direction was changed. After the direction was changed, the distance was advanced because there were no obstacles within the range of the mobile robot (Step.3). After traveling the distance , the direction was changed by -2θ to avoid obstacles (Step.4). After changing direction, the mobile robot went straight (Step.5) and went straight on the black line, which is the running path (Step.6). From Fig. 8, the mobile robot can detect obstacles in real time and avoid them without collision, and good results are obtained. It is considered that practical obstacle avoidance is possible by integrating the distance information and the width information of the obstacle.

Conclusions
Currently, the demand for indoor mobile robots is on the rise. Therefore, a system for detecting indoor walls is indispensable, and a highly accurate system is required. In the conventional method, a plurality of sensors are incorporated or an expensive sensor is used. However, this research system consists of a line laser and a camera. A lowcost, high-precision system was realized. Many mobile robots use basic cameras. This system can be implemented simply by installing a line laser on top of the mobile robot.
In this research, in order to allow the mobile robot to avoid obstacles, avoid collisions, and run without stopping, an approach using a line laser and a camera, which is different from the conventional obstacle detection method, is used for obstacles.
The method proposed in this study depends on the irradiated part of the line laser. The part not irradiated with the line laser cannot be detected. Improving this point is a future task. It is thought that a wider range of obstacles can be detected by using a camera with a wide angle and low distortion. In addition, in this system, experiments were conducted using one line laser, but it is possible to reduce the undetectable part by increasing the number of lasers with different irradiation angles.