With the development of social economy, the scale and quantity of large passenger flows such as supermarkets, airports, stations, convention and exhibition centers and logistics warehouses continue to expand. The high-level and high-end commercial advantages of large and medium-sized cities continue to increase, and the demand for safety automation is increasing. Coming more and more urgent. The patrol robot is a multifunctional integrated system that can perceive the environment, plan routes, make dynamic decisions, conduct behavioral control and accompany alarm devices. In order to ensure the safety of the place, the robot can be used to conduct fixed-point surveillance or uninterrupted patrol monitoring of the place. At present, the role of patrol robots in safety work is becoming more and more important, and its research has attracted more and more attention at home and abroad. It has become a new research direction of service robots. The application and development of police patrol robots are changing with each passing day. Various police patrol robots are widely used in security, fire protection, demolition, reconnaissance, anti-terrorism and other fields, which are of great significance to the maintenance of national public safety.
1 Introduction to Lidar
Lidar is a sensor that feels the external environment. The sensor refers to a detection device that can transform the information to be measured and output it according to certain laws. It is an important hardware to feel the external environment. It is mainly divided into vision, position, speed, force, and touch, such as high-speed cameras . Vision input systems, such as cameras and 3D lidars, laser scanners and other environmental vision input systems. In smart devices, sensors are equivalent to the sensory organs of devices such as driverless cars or robots, as the only autonomous input other than manually setting parameters. It determines the interaction capability between the device and the external environment and the accuracy and richness of device information input.
Lidar is a sensor based on non-contact laser ranging technology. It can detect targets by emitting laser beams to obtain accurate three-dimensional images. It can obtain high-precision contour information of the local environment in real time, and the ranging accuracy can reach centimeter level. It has the advantages of high accuracy, fast speed, and high efficiency. It is an indispensable part of navigation, surveying and mapping.
Table 1 Various lidar functions and application scenarios
Lidar is mainly divided into one-dimensional lidar, two-dimensional lidar and three-dimensional lidar. Different lidars are used in different environments (see Table 1).
At present, with the rapid development of artificial intelligence, the advantages of 3D lidar are becoming more and more prominent, and the scope of application is becoming wider and wider.
2. Working principle of 3D lidar
Lidar determines the geometric position of its surroundings by emitting laser pulses that reflect surrounding objects and return to the sensor. It calculates the time it takes for the pulse to reach the object, reflect and return to the sensor, and use this time to calculate the distance of the object. The scanner builds a complete point cloud by rotating the emitter (or a group of emitters) so that pulses are sent in multiple directions. Different lidar scanners differ in many aspects, from their accuracy, the area or volume they can scan, how much information they collect, and on this basis, how they must be installed to be useful for a particular application. Another consideration is which software tools can interface with the hardware and collect data in order to present or process the data in a useful way.
The core components of 3D laser navigation include 3D lidar, gyroscope and encoder. Its core technology is mainly map ratio structure and real-time matching and positioning of laser data. It first performs information fusion to solve the position, and then calculates the robot's global position coordinates through laser ranging, and realizes trackless navigation according to the coordinate position
When the robot enters the patrol area for the first time, it will use 3D lidar to scan the surrounding environment and generate an environment map through synchronous map construction and positioning algorithms. In the implementation of the inspection task, the 3D laser real-time scanning terrain and environmental terrain are accurately matched to calculate the global position information, as shown in Figure 1. In some special cases, the terrain features scanned by 3D lidar are susceptible to interference, for example, in some outdoor environments with sparse terrain features. In order to ensure the accuracy and reliability of navigation, police patrol robots use inertial navigation technology to modify the terrain matched by the 3D lidar, so as to achieve an understanding of the global environment. By measuring its acceleration and angular velocity and the feedback information of the unit fusion encoder, an accurate trajectory recursive motion model is generated, and finally the speed and position information of the robot is output. In order to ensure the reliability of robot navigation, it is complicated. In order to ensure the reliability of robot navigation and avoid 3D laser interference, the navigation system introduces redundant sensor data cross-diagnosis technology to isolate interference signals. The algorithm flow of 3D laser navigation is as follows: read 3D point cloud image, inertia and mileage data; rasterization, noise reduction and impurity processing; calculate initial image position; calculate robot posture estimation; calculate robot posture observation; calculate robot's best position And direction.
Figure 1 Working principle of 3D lidar
3. The advantages of 3D lidar in the navigation of police patrol robots
Three-dimensional laser navigation technology has relatively little dependence on the external environment, has high positioning accuracy, can measure a long distance, and can process a large amount of environmental information. When the surrounding environment of the robot does not change much, it can be relatively stable and has little impact on global information, thereby reducing the dependence on fixed references in the environment. The main advantages of 3D laser navigation are: high signal-to-noise ratio (SNR): the amount of point cloud information is 16 times that of a 2D laser with the same resolution. Position and posture information: For harsh road conditions, point cloud images in a unified coordinate system can be obtained, and accurate position information can be output; high positioning accuracy: In most cases, the position accuracy can reach ±2 cm, and the angle accuracy can reach ±1° , Avoid obstacle avoidance: It can detect the front three-dimensional area, completely avoid obstacles and prevent falling
4. Key technology of patrol robot
Mobile robot technology and safety monitoring functions are an important part of patrol robots. Mobile systems, sensors, controllers, alarms and communication equipment are its general components. The mobile mechanism can be made into walking form, wheel type, crawler type and hybrid type. It is the main body of the patrol robot and determines the movement space and activity of the patrol robot. The sensing system generally uses ccD camera sensing system, ultrasonic rangefinder, laser rangefinder, touch and proximity sensor, infrared sensor, etc. In recent years, due to the rapid development of communication technology, computer technology, artificial intelligence technology and sensor technology, the research and development of patrol robots has also entered a period of rapid development and ushered in a relatively bright development prospect. The working environment of patrol robots is often uncertain and changes with time, so its safety, reliability and anti-interference must be considered. Therefore, multi-sensor information fusion technology is used to detect the environment to increase the complementarity and fault tolerance of information. The current key technologies include multi-sensor information fusion technology, navigation technology, and intelligent control. The following is a detailed description of navigation technology.
The navigation of the patrol robot is based on pre-given safety tasks, global path planning based on known map information, and continuous perception of local working environment information, and make various decisions on its own during travel. Adjust your posture and posture at any time to avoid obstacles and guide yourself to drive safely. At the same time, complete the safety work of timing, fixed-point or circular inspection. The working environment of the robot can be divided into static environment, dynamic known environment and dynamic uncertain environment. The research content of path planning can be divided into model-based path planning and sensor-based path planning according to the different ways the robot obtains environmental information. According to the robot's understanding of environmental information, it can be divided into global path planning and completely unknown or partially unknown environmental information. The working environment of the robot can be detected online through sensors. Local path planning for obtaining information about the location, shape and size of obstacles. With the increase in the number of application scenarios and the increasingly complex application environment, the navigation technology requirements of police patrol robots are getting higher and higher. The existing navigation methods mainly include magnetic stripe, laser, GPS, etc. These methods have obvious defects in the adaptability and stability of the environment. 3D laser navigation technology can effectively compensate for these shortcomings. It is less dependent on fixed reference objects and is insensitive to small changes in the surrounding environment. Therefore, 3D laser navigation technology can improve the adaptability and stability of the robot environment.
The key of the navigation system is to solve self-positioning, target planning and navigation planning. The map creation and positioning of police patrol robots are the basis of their navigation research and an important guarantee for the realization of "autonomy". Global path dynamic planning is one of the most important tasks in police patrol robot navigation. The main requirement of the police patrol robot path planning system is to find a path in the environment map. Ensure that the police patrol robot does not collide with the outside world when moving along the path; it can deal with the uncertain factors in the sensor-perceived environment model and the error in the path execution, and minimize the impact on the robot sensor by making the police patrol robot avoid external objects The impact of the sensing range. You can find the best path according to your needs. Common path finding algorithms include two categories: routing algorithms based on static data and routing algorithms based on real-time data . The static data routing algorithm is a method of calculating the optimal path based on historical data under the condition that the map network environment remains unchanged. The static data routing algorithm is very simple, but it requires one calculation to get the optimal path. However, the algorithm lacks real-time information collection and feedback, so it is not suitable for application in complex and changeable environments.
5. Application of 3D Lidar in the navigation of police patrol robots
During major festivals, there will be police patrol robots on duty to patrol crowded and important places. At first, 2D laser navigation technology was widely used, but because 2D laser navigation technology is highly dependent on the environment, these places are generally large in size, less marked, and densely populated, so the navigation effect is poor, and there is often a problematic position. 3D laser navigation technology is mainly used to solve this problem. It uses vertical size information to greatly improve the stability and accuracy of navigation. The 3D point cloud is used to construct a map of the scene environment. The ground, fences, safety sheds, crowds and trees that it navigates are clearly visible. The robot uses this rich data to accurately locate its position in the environment, so as to obtain more information about the surrounding environment. When the surrounding environment of the police patrol robot changes slightly, the impact on the overall situation is significantly reduced, thereby reducing the dependence on a fixed reference in the environment. It reduces positioning errors caused by changes in the surrounding environment, and greatly improves the stability and accuracy of navigation.