Shulou(Shulou.com)06/03 Report--

The menacing epidemic in the new areas has once again affected the hearts of the people of the whole country. We must maintain confidence, remain vigilant and win this epidemic prevention war! The addition of robots is bound to be the icing on the cake. In the quarantine zone, the robot realizes contactless distribution and ensures the life of the people in the quarantine area. In the hospital infection area, mobile robots can replace medical staff to enter the infection area to carry out disinfection, nursing, temperature measurement, cleaning and other work, reducing the risk of infection of medical staff. In the residential community, the robot realized epidemic prevention publicity and migrant population registration, ensuring people's normal life during the epidemic.

Equipped with the distribution, disinfection and inspection robot of Silan scheme.

The realization of these functions is inseparable from the intelligent mobile ability of the robot. In the previous article, we mentioned the significance of chassis to mobile robots, which can help manufacturers quickly integrate and realize landing applications.

As we all know, autonomous positioning and navigation includes positioning, mapping and path planning. When it comes to positioning and navigation, it is bound to think of SLAM, but SLAM only completes the positioning and map creation of these two things, it is not completely equivalent to autonomous positioning and navigation.

Positioning

Whether you have a map or not, you must know your location before you go to your destination, and so do robots. However, our positioning depends on our eyes, while robots rely on "lidar."

Lidar scanning profile data

Building a map

When we live in an unfamiliar environment, navigation software and outdoor maps become the most favorable tools for us to find our way. Service robots, like human beings, also need to rely on maps to perceive the external environment. Different robots that differ from sensors through algorithms will take different forms of map description.

There are three main processes in building SLAM maps:

(1) preprocessing: optimize the original data of the point cloud formed by radar, eliminate some problematic data, or filter.

(2) matching: the point cloud data of the current local environment are matched by finding the corresponding position on the established map.

(3) Map fusion: a new round of data from lidar is spliced into the original map, and finally the map is updated.

The process of map preprocessing, matching and fusion

At present, raster map is the most widely used map storage method for robots.

Grid map is to divide the environment into a series of grids, in which each grid is given a possible value, indicating the probability of the grid being occupied, and each "pixel" represents the probability distribution of obstacles in the actual environment.

The formation of raster map

The process doesn't sound complicated, but it still encounters a lot of unknown problems.

Circular corridor closure and disconnection

In early 19 years, Silan Technology launched SLAM 3.0 to deal with this problem. When the robot moves to the original environment that has been explored, SLAM 3.0 can rely on the internal topology diagram for active closed-loop detection. When the new closed-loop information is found, SLAM 3.0 uses algorithms such as Bundle Adjuestment (BA) to modify the original pose topology map (that is, map optimization), so as to effectively modify the post-closed-loop map and achieve more reliable environmental mapping.

Closed loop correction

Path planning and motion control

When positioning and mapping are done, the next problem to be solved is the ability to navigate between target points An and B.

Path planning is divided into global planning and local planning.

Global planning: it is the top-level motion planning logic, which finds the quickest path to the target point on the map according to the pre-recorded environment map of the robot and combining the current position of the robot and the location of the task target point.

Local planning: when the environment changes or the upper-level planned path is not conducive to the actual walking of the robot (for example, the robot can not complete a specific turning radius according to the planned path), the local path planning will be fine-tuned.

Hierarchical motion planning framework and corresponding output data

With the cooperation of these two levels of planning modules, the robot can realize the intelligent movement from point A to point B. However, in the actual working environment, the above configuration is not enough. Because the process of motion planning also includes static map and dynamic map.

A * algorithm

A* (A-Star) algorithm is not only the most effective direct search method to solve the shortest path in static road network, but also an effective algorithm to solve many search problems. The closer the estimated distance value in the algorithm is to the actual value, the faster the final search speed is. However, the A* algorithm can also be used in dynamic path planning, but the route needs to be replanned when the environment changes.

D* algorithm

D* algorithm is a mainstream algorithm at present, the biggest advantage is that there is no need to detect the map in advance, the robot can act like people, even in an unknown environment, and the path will be adjusted all the time as the robot continues to explore.

The above algorithms are the path planning algorithms needed by most robots at present, which can make the robot as intelligent as human, quickly plan the shortest path from point A to point B, and know how to deal with obstacles. However, as one of the earliest service robots in the consumer market, sweeping robot needs a somewhat different path planning algorithm.

Therefore, for robots in different scenarios, autonomous positioning and navigation technology needs to be constantly updated to support more scene applications. Only when the navigation technology is done, can the whole robot make a qualitative leap, and the layout of the robot from 0-1 can be faster and faster.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.