Nowadays, laser navigation sweeping and mopping robot are becoming more and more popular. More and more families want to buy a sweeping robot. The sweeping robot has become a kind of "new small household appliances". In order to achieve more efficient and energy-saving cleaning of the whole house, the navigation system of the sweeping robot is an indispensable function. Therefore, the robot vacuum manufacturer will spend a lot of energy preparing a pair of "golden eyes" for the robot vacuum, allowing the robot vacuum to "see" the world and plan the path according to the world in the "eye".

For now, there are two widely used navigation technologies for sweeping robots on the market. One is laser navigation and the other is visual navigation. So which one is better?

Laser navigation:

Sweeping robots equipped with laser navigation, such as Mijia LDS sweeping and dragging robots, use LDS lidar scanning to determine their position. The laser is emitted through high-speed rotation, and then the distance between oneself and the obstacle is judged by the time when the laser is emitted and the obstacle is reflected, so as to judge the relative position and realize positioning.

The advantage of this navigation method is that it has high positioning accuracy, even in a non-light environment. Laser navigation, with its precise positioning advantages, occupies most of the market share of sweeping robot navigation products. But it also has obvious shortcomings. Its main sensors are mostly located on the top of the sweeping robot, and the viewing angle is very small in the vertical direction, which cannot realize large-scale obstacle detection. For obstacles that are lower or slightly higher than the fuselage (higher than the fuselage, but will encounter the lidar sensor), such as the edge of the cabinet bottom and the wires near the ground, the sweeping robot has a poor obstacle avoidance effect. Therefore, problems such as the sweeping machine being stuck at the bottom of the cabinet and being entangled by the wire body often occur.

Visual navigation:

Sweeping robots equipped with visual navigation, such as Cobos DJ65, use monocular vision navigation to achieve positioning. The vision sensor of this sweeping robot is installed on the top, and through a complex algorithm, the robot can be positioned by perceiving an optical image composed of light spots with different brightness. These optical images look different from different angles. By continuously collecting the image information, the robot can locate on the map built by itself, so as to know which positions have been scanned and which places need to be cleaned.

The advantages of visual navigation are low cost, strong applicability, and technical routes can be solved with the help of big data. However, the problem of the visual navigation sweeper is also obvious. The premise of visual ranging is accuracy. After the camera obtains the environmental information, the distance must be calculated quickly. If the data processing and algorithms cannot keep up during the period, the distance data will be biased.

​

Therefore, from the perspective of technical realization, visual navigation is undoubtedly much more difficult than laser navigation. Because of this, there are very few sweeping robots on the market that fully adopt a vision navigation system, and more are laser navigation + vision sensors or other sensor fusion solutions.

From the perspective of future development, regardless of industry development or product user experience, the potential of visual navigation is beyond doubt. In addition to the leading gooses in the sweeping robot industry such as Cobos and iRobot, there are still many outstanding domestic companies that continue to "conquer" problems in the field of visual navigation. For example, the domestic stereo vision solution provider INDEMIND has launched a vision navigation and positioning solution for sweeping robots.

It is understood that the INDEMIND visual navigation solution uses a binocular stereo vision architecture, and the program has a high degree of maturity, and has established cooperation with many companies in the industry. This solution uses stereo vision technology and high-precision Vi-SLAM algorithm to integrate the positioning and interaction problems of sweepers that should have been involved in the ultra-long industrial chain into a set of solutions to quickly provide customers with high-precision navigation and positioning and paths. Core functions such as planning, intelligent obstacle avoidance, goal following, and home recognition. It is worth mentioning that the navigation accuracy of this solution has achieved the leading level of absolute positioning accuracy <1% and attitude accuracy <1°, which is comparable to lidar.

In addition, due to the natural advantages of the vision solution, the INDEMIND vision navigation and positioning solution can control the navigation and positioning cost of the sweeping robot at the hundred yuan level, which greatly reduces the manufacturer's R&D and production costs.

In general, compared with lidar solutions, visual navigation is a navigation and positioning solution for sweeping robots with higher dimensions, lower cost, and stronger applicability. In the future, with the increase in computing power and the reduction in computing power costs, the cost of visual navigation solutions can be further explored; with the in-depth research, the environmental semantic functions brought by visual navigation will become more and more perfect, and intelligent sweeping robots will become more visualized , Smarter, and the price is more affordable.

The team of Guangdong Anjiu Intelligent Technology Co., Ltd. has been committed to providing overall technical solutions for intelligent sweeping robots for well-known brands such as Haier for many years. The team has accumulated a good reputation in the industry; technical strength and development capabilities have been recognized by many international brands and live in smart The forefront of AI development. For more information about laser navigation sweeping and mopping robots, please follow Anjiu Information.