WO2022017296A1

WO2022017296A1 - Robot motion information recognition method, obstacle avoidance method, robot capable of obstacle avoidance, and obstacle avoidance system

Info

Publication number: WO2022017296A1
Application number: PCT/CN2021/106897
Authority: WO
Inventors: 钟扬
Original assignee: 炬星科技（深圳）有限公司
Priority date: 2020-07-23
Filing date: 2021-07-16
Publication date: 2022-01-27
Also published as: CN111844038B; CN111844038A

Abstract

A robot motion information recognition method, an obstacle avoidance method, a robot capable of obstacle avoidance, and an obstacle avoidance system. The method comprises: on a current travel path of a robot, generating a first projection pattern carrying robot motion information; visually capturing the first projection pattern; and recognizing the robot motion information carried by the first projection pattern. By transmitting robot motion information by means of a projection pattern, visually capturing the projection pattern for recognition, and further performing obstacle avoidance planning, the robot is allowed to obtain travel-related information of other robots on the route before the lidar recognizes other robots, so as to avoid the same in advance; one the one hand, safety can be improved and the probability of collisions at corners can be reduced; on the other hand, operating speed can be increased while improving the safety, and efficiency can be improved.

Description

A robot motion information recognition method, obstacle avoidance method, obstacle avoidance robot, and obstacle avoidance system

technical field

The invention belongs to the technical field of robot obstacle avoidance, and particularly relates to a robot motion information identification method, an obstacle avoidance method, an obstacle avoidance robot, and an obstacle avoidance system.

Background technique

During the operation of AMR (Autonomous Mobile Robot), it is necessary to identify obstacles and then trigger obstacle avoidance, including actions such as stopping and detouring, to avoid obstacles and prevent collisions. These obstacles include fixed obstacles such as walls and shelves, as well as moving obstacles such as pedestrians. Generally speaking, AMR uses lidar to identify the outline of objects to avoid obstacles. When it detects that the obstacle is close to itself within a certain distance, it triggers obstacle avoidance. The longer the obstacle avoidance distance is, the safer it is but the more efficiency is lost.

Other robots are also high-speed moving objects in the scene and are one of the main objects for obstacle avoidance. Since robots move faster than pedestrians and not as flexible as pedestrians, there is a greater chance of collisions between robots. For this reason, it is necessary to reduce the running speed of the robot and increase the obstacle avoidance distance, which leads to the loss of efficiency. At the same time, in some scenarios that are prone to collision due to occlusion, such as narrow intersections (one robot cannot detect robots on the other side), the robot has a higher chance of colliding.

technical problem

The present invention provides a robot motion information identification method, an obstacle avoidance method, an obstacle avoidance robot, and an obstacle avoidance system, so as to solve the problems of collision and crowding easily caused by the multi-robot cooperative movement in the prior art, and make the robot more convenient when moving. Flexible and safe, avoid collisions and blockages in complex scenarios such as narrow spaces or intersections, and improve the efficiency of multi-robot collaboration.

technical solutions

In a first aspect, the present invention provides a method for identifying motion information of a robot, including:

On the current travel path of the robot, a first projection pattern carrying the motion information of the robot is generated;

visually capturing the first projected pattern;

Identify the robot motion information carried by the first projection pattern.

In a second aspect, the present invention also provides a multi-robot motion obstacle avoidance method, comprising: the first robot projects a first projection pattern carrying its own motion information on the ground along its travel direction;

The second robot captures the first projection pattern through a vision device installed on its body;

The second robot recognizes the motion information of the first robot corresponding to the first projection pattern, and performs corresponding obstacle avoidance measures in combination with its own motion information to avoid collision with the first robot.

In a third aspect, the present invention also provides a robot that can dynamically avoid obstacles, including:

The main body of the robot can be moved autonomously;

a projection device, which is relatively fixedly arranged on the robot body; the projection device includes an electronic display and a plurality of projection films for the robot to project different projection patterns;

A vision device, which is installed on the robot body on the same side as the projection device, and is used to capture the projection pattern of the robot's travel area.

In a fourth aspect, the present invention also provides a multi-robot obstacle avoidance system, including:

a number of robots that move autonomously in the same work environment;

a projection module, in response to the autonomous movement of the robot, generating different projection patterns in the travel path of the robot, the projection patterns carrying the movement information of the robot;

a vision module for capturing the projected pattern in the travel path of the robot;

The processing module is used for identifying the motion information of the robot corresponding to the projection pattern, and combining the motion information of the other robot to control the other robot to perform corresponding obstacle avoidance measures.

beneficial effect

The invention provides a robot motion information identification method, an obstacle avoidance method, an obstacle avoidance robot, and an obstacle avoidance system. On the current travel path of the robot, a first projection pattern carrying the motion information of the robot is generated, the first projection pattern is visually captured, and the motion information of the robot carried by the first projection pattern is recognized. Among them, the motion information of the robot is transmitted through the projection pattern, and the visual capture projection pattern is used for identification processing, and obstacle avoidance planning is further adopted, so that the robot can obtain the driving information of other robots on the route before the lidar recognizes other robots. Pre-evasion can improve safety and reduce the probability of corner collision on the one hand; on the other hand, it can improve the running speed and increase efficiency on the basis of safety.

Description of drawings

FIG. 1 is a flowchart of an embodiment of a method for recognizing motion information of a robot according to the present invention.

FIG. 2 is a flowchart of another embodiment of the motion information identification method of the present invention.

FIG. 3 is a flowchart of an embodiment of the multi-robot dynamic obstacle avoidance method of the present invention.

FIG. 4 is a flowchart of another embodiment based on the embodiment of FIG. 3 .

FIG. 5 is a flowchart of another embodiment of the multi-robot dynamic obstacle avoidance method of the present invention.

FIG. 6 is a schematic structural diagram of a device of the obstacle avoidance robot of the present invention.

FIG. 7 is a schematic diagram of another device structure of the obstacle avoidance robot of the present invention.

FIG. 8 is a schematic diagram of projection and visual capture of the obstacle avoidance robot of the present invention.

FIG. 9 is a schematic diagram of the obstacle avoidance method of the obstacle avoidance robot of the present invention.

FIG. 10 is a schematic block diagram of the robot obstacle avoidance system of the present invention.

Embodiments of the present invention

In order to make the objectives, technical solutions and advantages of the present disclosure more clear, the present disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention. The technical features involved in the different embodiments of the present disclosure described below can be combined with each other as long as there is no conflict with each other.

As shown in FIG. 1 , FIG. 1 is a schematic flowchart of an embodiment of a method for recognizing motion information of a robot of the present invention; a method for recognizing motion information of a robot of the present invention may be implemented as steps S10 - S30 described below.

Step S10, generating a first projection pattern carrying the motion information of the robot on the current travel path of the robot.

In this embodiment, the working scene of the robot is generally a large area, such as inside various types of warehouses. When the robot receives a task, it will go to the task destination point according to the task prompt, and the robot will go from the starting point to the destination point according to the map. A rough motion path is generated, the robot travels along the path, and a first projection pattern is generated on the ground in front of the robot. The first projection pattern follows the real-time motion of the robot and corresponds to the current motion state of the robot. Specifically, the preset distances between the first projection pattern and the robot are different distances such as 2 meters, 5 meters, and 10 meters. It can be understood that the greater the movement speed of the robot, the longer the projection distance. The motion information of the robot includes the robot's own code, speed, driving direction, distance from the projected pattern, etc. These information can be represented by different projected patterns. The projected pattern is a pattern that the robot can recognize its semantics, such as QR (Quick Response) code II. dimensional code form.

In this embodiment, several preset motion states that the robot prefers can be used as the basic projection pattern. Go straight, stop and other conventional steering, and then take several projection distance values such as 2 meters, 3 meters, 5 meters, 10 meters, etc., extract the values from the above attributes, and add the robot's own code to form the basic projection pattern. Further, also The projection distance and speed can be integrated into a time attribute, that is, the robot keeps its motion state unchanged, and the time interval between reaching the first projection pattern, such as 3s, 5s, etc., means that the robot will reach the first projection generated at this moment after 3s pattern. The above basic projection patterns are taught to the robot in advance, so that in the actual operation scene, the robot can quickly identify the projection pattern, and make corresponding planning measures to avoid obstacles.

Among them, in the projection pattern recognition method, a recognition model can be generated by machine training using pre-prepared supervision data, and image recognition can be performed by using the obtained recognition model. Further, the projection pattern is divided according to the number of robot motion attributes. For example, the projection pattern is composed of four parts: robot number, direction, speed and projection distance. The orientation of each part in the projection pattern is preset, and a small image of a fixed size is used. Machine training of a regressor that takes as input and outputs projected identification codes in small images. Then, after the recognition object image is divided into sub-regions according to the corresponding preset, each sub-region is adjusted to a fixed size to generate a small image, and the trained regressor obtains the motion attribute value in each small image.

Step S20, visually capturing the first projection pattern.

In this embodiment, the first projection pattern is captured visually, that is, a camera, a visual sensor and other devices are used for snapshot or real-time monitoring to obtain the image information of the first projection. It can be understood that the generated first projection pattern is The visual device can easily capture the image presentation form. Further, the visual device can be set to capture a preset pattern type and then upload, store and other processing. If a conventional ground image is detected, no further processing is performed.

It can be understood that, in this embodiment, the visual capture of the first projection pattern may be that the first robot itself captures the first projection pattern, or the second robot captures the first projection pattern, or other visual objects in the working environment. device to capture. When the first robot captures the first projection pattern with its own motion state, the first robot can do nothing by default or confirm whether the projection pattern is displayed normally on its travel path. Abnormal feedback; when the second robot captures the first projection pattern, in this embodiment, the second robot refers to any robot except the robot corresponding to the information carried by the first projection pattern, and the second robot recognizes the first projection pattern, Obtain the motion information of the first robot, and then understand the motion speed, motion direction, and position of the first robot from the first projection pattern, and determine the approximate position where it meets the first robot based on its own motion speed and direction, and Calculate the collision probability, if the collision probability is greater than the preset value, take obstacle avoidance measures, such as deceleration, acceleration, stop, steering and other actions.

Step S30, identifying the robot motion information carried by the first projection pattern.

In this embodiment, a processing device may be set up in the working environment to identify the first projection pattern and perform further processing. Schedule robots in this area. In another embodiment, a processing device is provided on the robot body, and the processing device can be used alone as a sub-processing system of the robot, or the processing device can be integrated or embedded in the robot central processing system in the form of a processing module for identification. The first projection pattern, and unified integration processing and control.

Based on the description of the embodiment shown in FIG. 1 and in conjunction with FIG. 2 , step S10 further includes:

Step S11 , the first robot projects the first projection pattern on the ground along its travel direction or/and the first projection pattern is projected by an external projection device in the working environment of the first robot.

In this embodiment, the first robot refers to any robot in the robot cluster in the same working environment or in the same space and on the same map. When the robot receives a task, each robot will form its own path plan. It travels along its own planned path, and on the path it travels, that is, right in front of the robot's walking direction, a first projection pattern is generated, that is to say, the generation position of the first projection pattern is the position the robot is about to reach, and the first projection pattern is generated. The projection pattern moves with the first robot, and the distance between the first projection pattern and the robot can be preset to be different distances such as 2 meters, 5 meters, and 10 meters.

The first projection pattern is generated by the projection of the first robot itself, and the generated first projection pattern carries the motion information of the first robot. The motion information of the robot includes the robot's own code, speed, driving direction, distance from the projection pattern, etc. These information can be It is represented by different projection patterns, and the projection pattern is a pattern that the robot can recognize its semantics, such as the form of QR (Quick Response) code.

In another embodiment, an external projection device in the first robotic work environment projects the first projection pattern. Preferably, the external projection device performs the projection operation on some preset road sections, which are intersections or areas where robots move frequently. Further, the projection device is further provided with a detection device for detecting a robot about to enter the projection road section, and projecting the detected robot motion information to a preset projection area to generate a first projection pattern.

Optionally, the projection device is provided with a communication device for receiving motion information sent by a robot about to enter the projection area, and projecting a projection pattern corresponding to the information to a preset projection area. It can be understood that the projection device can also perform dynamic projection following the movement direction of the trolley in the projection section, that is, the projection pattern and the trolley move in unison.

In this embodiment, several motion states that the robot prefers can be preset as the basic projection pattern. For example, in an embodiment, the first projection pattern only contains two pieces of robot motion information: the movement direction and the arrival time. The arrival time is The robot keeps its motion state unchanged, and the time interval between reaching the first projection pattern, such as 3s, 5s, etc., means that the robot will reach the first projection pattern generated at this moment after 3s. In this example, set the movement direction to be straight , left, right, the arrival time is 3s, 4s, 5s, then the basic projection pattern has 3, 4, 5s straight, 3, 4, 5s to the left, 3, 4, 5s to the right, a total of 9 types. 9 kinds of projection pattern numbers 01-09, when the robot is about to enter the projection area, it only needs to send the number between 01-09 to the projection device, and the projection device will project the corresponding projection pattern to the projection section.

Based on the description of the embodiment shown in FIG. 1 and in conjunction with FIG. 2 , step S20 further includes:

Step S21, the second robot captures the first projection pattern through a vision device installed on its body; or/and a vision device installed in the robot operating environment to capture the first projection pattern.

In this embodiment, the second robot refers to any other robot except the robot corresponding to the information carried by the first projection pattern. A vision device is installed on the second robot body, and the vision device is a camera, a vision module, or a vision sensor, and is used for snapshot or real-time monitoring to obtain the image information of the first projection.

It can be understood that the generated first projection pattern is an image presentation form that the visual device can easily capture. Further, the visual device can be set to capture the preset pattern type and then upload, store and other processing, or preset the second robot to enter. Go to the projection area and turn on the vision device to capture and feed back the image to the robot processing system in real time.

In another embodiment, a vision device installed in the robotic work environment captures the first projected pattern. Further, the vision device captures the projection pattern in the working environment of the robot, and sends the captured projection pattern to the processing module of the robot, and the processing module of the robot recognizes the projection pattern information, makes judgments, and takes obstacle avoidance measures. The equipment can be scattered at the intersection, and a visual device is only responsible for capturing the projection pattern of one intersection, and only communicates with the robot in the intersection section, and the other second robot in the intersection section receives the first projection pattern and transmits it. Identify and process, and take corresponding obstacle avoidance measures in combination with its own motion state.

In another application scenario of this embodiment, the vision device captures the projection patterns of the entire robot operating environment, that is, all projection patterns on the robot walking map, and feeds back all projection pattern information to the robot management system platform for processing and real-time monitoring by the platform. The robot's motion status is analyzed, and the robot's motion in high-risk road sections is analyzed, and obstacle avoidance instructions are given to robots with a higher collision risk.

Based on the description of the embodiment shown in FIG. 1 and in conjunction with FIG. 2 , step S30 further includes:

Step 31: The processing device installed on the robot receives the first projection pattern, and identifies the motion information of the first robot corresponding to the projection pattern; or/and the processing device installed in the robot operating environment receives the first projection pattern. A first projection pattern, identifying motion information of the first robot corresponding to the first projection pattern.

In this embodiment, the processing device on the robot receives the first projection pattern, and identifies motion information of the first robot corresponding to the first projection pattern. Optionally, the processing device is separately installed on the robot body, as an independent sub-processing system, used to complete the dynamic obstacle avoidance of the robot. It can be understood that the processing device can also be set as a processing module integrated on the robot central processing unit. , the second robot recognizes the first projection pattern through the processing device, obtains the movement information of the first robot, and then understands the movement speed, movement direction of the first robot, and the position away from the first projection pattern, combined with its own movement speed and direction, Determine the approximate position where it meets the first robot, and calculate the collision probability. If the collision probability is greater than the preset value, it will take obstacle avoidance measures, such as deceleration, acceleration, stop, steering and other actions.

In another embodiment, the processing device installed in the robot working environment receives the first projection pattern, and identifies the motion information of the first robot corresponding to the first projection pattern. Preferably, the processing device is arranged in one or more projection areas, and the projection areas are some preset road sections, and the road sections are intersections or areas where robots move frequently.

Further, the processing device is provided with a communication device for receiving the projection pattern image information sent by the vision device in the projection area, identifying the motion information of the robot, and calculating the collision probability in combination with the motion state information of other robots in the projection area, such as the collision probability. If it is greater than the preset value, it will instruct the obstacle avoidance measures, such as sending deceleration, acceleration, stop, steering and other action commands to the corresponding other robots in the projection area. Optionally, the processing device may also be a server for uniformly processing the projection pattern information of all projection areas in a warehouse and making obstacle avoidance planning.

As shown in FIG. 3 , FIG. 3 is a schematic flowchart of a multi-robot motion obstacle avoidance method according to the present invention. A multi-robot motion obstacle avoidance method of the present invention may be implemented as steps S100-S300 described below.

Step S100, the first robot projects a first projection pattern carrying its own motion information on the ground along its travel direction;

Step S200, the second robot captures the first projection pattern through a visual device installed on its body;

Step S300, the second robot recognizes the motion information of the first robot corresponding to the first projection pattern, and performs corresponding obstacle avoidance measures in combination with its own motion information to avoid collision with the first robot.

10, in step S100 in this embodiment, the first robot 100 refers to any robot in the robot cluster in the same work environment or in the same space and in the same map. When the robot receives a task, each robot The robot will form its own path plan. The robot travels along its own planned path. On its travel path, that is, directly in front of the robot's walking direction, a first projection pattern is generated, that is, the generation position of the first projection pattern. It is the position that the robot is about to reach. The first projection pattern moves with the first robot. The distance between the first projection pattern and the robot can be preset as 2 meters, 5 meters, 10 meters and other different distances.

The first projection pattern is projected and generated by the first robot 100 itself, and the generated first projection pattern carries the motion information of the first robot. The motion information of the robot includes the robot's own code, speed, driving direction, distance from the projection pattern, etc. These information It can be represented by different projection patterns. The projection pattern is a pattern that the robot can recognize its semantics, such as the form of QR (Quick Response) code.

In step S200 in this embodiment, the second robot 200 refers to any other robot except the robot corresponding to the information carried by the first projection pattern. A vision device is installed on the body of the second robot 200 , and the vision device is a camera, a vision module, or a vision sensor, and is used for snapshot or real-time monitoring to obtain the image information of the first projection.

It can be understood that the generated first projection pattern is an image presentation form that can be easily captured by the visual device. Further, the visual device can be set to capture a preset pattern type and then perform processing such as uploading, storing, etc., or preset the second robot 200. Enter the projection area and turn on the vision device to capture and feed back the image to the robot processing system in real time.

In one embodiment, in step S300 in this embodiment, the processing device on the second robot 200 receives the first projection pattern, and identifies motion information of the first robot corresponding to the first projection pattern. Optionally, the processing device is separately installed on the robot body, as an independent sub-processing system, used to complete the dynamic obstacle avoidance of the robot. It can be understood that the processing device can also be set as a processing module integrated on the robot central processing unit. , the second robot recognizes the first projection pattern through the processing device, obtains the movement information of the first robot, and then understands the movement speed, movement direction of the first robot, and the position away from the first projection pattern, combined with its own movement speed and direction, Determine the approximate position where it meets the first robot, and calculate the collision probability. If the collision probability is greater than the preset value, it will take obstacle avoidance measures, such as deceleration, acceleration, stop, steering and other actions.

As shown in FIG. 4, combined with the embodiment shown in FIG. 3, FIG. 4 is another embodiment of the multi-robot obstacle avoidance method of the present invention, and step S300 further includes:

Step S310, the second robot 200 is based on the optimal mutual collision avoidance algorithm, according to the current position and speed of the first robot 100 or/and the position and speed of the first robot 100 in the first projection pattern, Perform obstacle avoidance planning and change its speed direction; or the second robot 200 calculates the first time when the first robot 100 reaches the first projection pattern, and changes its movement speed according to a preset time interval to make It reaches the first projection pattern earlier or later by the time interval than the first time.

In an example of this step S310, based on the optimal mutual collision avoidance algorithm, the robots perform obstacle avoidance planning according to the current position and speed of the first robot 100 or the position and speed of the first robot 100 in the first projection pattern , change its speed and direction. In this embodiment, the optimal mutual collision avoidance algorithm is an ORCA (Optimal Reciprocal Collision Avoidance) obstacle avoidance algorithm to navigate the projection area. It can be understood that, in this embodiment, the first robot 100 and the second robot 200 are opposite, that is to say, the second robot 200 can also project its own projection pattern, and the first robot 100 can also capture and recognize the projection pattern, the robot in this scenario is the first robot 100 and the second robot 200 . Specifically, the second robot 200 obtains the speed and current position of the first robot 100 by identifying the projection pattern information, and the first robot 100 also obtains the speed and current position of the second robot 200 by identifying the projection pattern information. Adopt the same obstacle avoidance strategy and jointly select a new speed to avoid obstacles.

In another embodiment, the second robot 200 calculates the first time when the first robot 100 arrives at the first projection pattern, and changes its movement speed according to a preset time interval to make it advance or delay the first time by a certain amount. The time interval reaches the first projection pattern. In this embodiment, the preset time interval is 1 s, 2 s or 3 s, etc. Further, the second robot 200 determines which robot reaches the projection at the moment first without changing the motion state of the second robot 200 in combination with its own motion information. pattern, if the first robot 100 reaches the projection pattern first or at the same time, then control itself to decelerate until it reaches the projection pattern after a preset time delay, such as 2s, than the first robot 100. If it is judged that the second robot 200 reaches the projection pattern at this moment first, then Control itself to accelerate until it reaches the projected pattern after a preset time such as 2s ahead of the first robot 100 . It can be understood that, in this embodiment, as long as the second robot 200 takes obstacle avoidance measures, the first robot 100 can maintain its own motion state to perform operations. A possible implementation example: According to the working space of the robot and the frequency of the road sections that the robot travels, the path traveled by the robot is divided into main roads and secondary roads. The main road is a road section with few robots or a road section that performs secondary tasks. The robot on the main road is the first robot 100, which can only project the projection pattern without capturing the projection pattern. The robot on the secondary road is the second robot 200, which only captures the projection pattern. The pattern does not project a projected pattern, in other words, the first robot 100 drives on the main road, the second robot 200 drives on the secondary road, the second robot 200 actively avoids obstacles, and the first robot 100 can maintain its own desired speed. , to complete more tasks.

Based on the description of the embodiment shown in FIG. 3 , as shown in FIG. 5 , FIG. 5 is a schematic flowchart of another implementation manner of a multi-robot motion obstacle avoidance method. In the embodiment shown in FIG. 5 , in “step S300 of the embodiment shown in FIG. 2 , the second robot identifies the motion information of the first robot corresponding to the first projection pattern, and performs corresponding motion information in combination with its own motion information. obstacle avoidance measures to avoid collision with the first robot", and then step S400 is also executed.

Step S400 , the second robot projects a second projection pattern on the ground along its travel direction after the obstacle avoidance measure.

In this embodiment, the second projection pattern represents the motion information of the second robot 200 after making the obstacle avoidance plan. For example, after the second robot 200 recognizes the projection pattern of the first robot 100 and combines its own motion state, it is calculated to obtain There is a high possibility of collision, and obstacle avoidance measures need to be taken to reduce the original speed of 2m/s to 0.5m/s. When the speed reduction is completed, the second robot 200 is on the current travel path, that is, the travel after obstacle avoidance planning. path to generate a second projection pattern.

It should be noted that the projection pattern is not updated in real time according to the motion state of the robot, that is, during the process of decreasing from 2m/s to 0.5m/s, the projection pattern remains unchanged, and the projection is updated only after the obstacle avoidance planning action is completed. pattern, that is, changing from the first projection pattern to the second projection pattern; on the other hand, when the robot encounters a static obstacle, it may temporarily change its motion state, such as slowing down and passing through. In this static obstacle avoidance process, although The change of decelerating first and then returning to the original speed is also completed, but the projection pattern may not be changed to keep the first projection pattern. In this way, the number of times of transformation of the projection pattern can be reduced, and at the same time, the pattern of the projection pattern will not be too many, the difficulty of identification processing is reduced, the rapid identification and operation of the robot is facilitated, and it is easier to meet the performance requirements of the projection device and improve the practicability. .

As shown in FIG. 6 , the present embodiment discloses a device structure of an obstacle avoidance robot, and the robot includes:

The robot body 90 can be moved autonomously, and the robot body 90 has complete functions of an autonomous mobile robot such as AMR (Autonomous Mobile Robot).

The projection device 60 is relatively fixedly arranged on the robot body; the projection device 60 includes an electronic display and several projection films for the robot to project different projection patterns.

The vision device 70, which is installed on the robot body on the same side as the projection device 60, is used to capture the projection pattern of the robot's travel area.

In this embodiment, the projection device 60 can automatically switch the projection films to project different projection patterns, and each projection pattern represents different robot motion information. Specifically, the projection device 60 includes a plurality of projection films. The projection angle of one or more projection light sources to switch the use of slides.

Optionally, more projection patterns can be generated by combining patterns and colors. The projector includes a laser light source 1R of various colors, a laser light source 1G, a laser light source 1B, a collimator lens, and a lenticular lens. lens), spatial modulation elements, projection lenses and dichroic mirrors. The laser light sources 1R, 1G, and 1B emit red, blue, and green laser light in this order. After passing through the collimator lens, the green laser beam becomes substantially parallel light, which is reflected by the mirror and then transmitted through the dichroic mirror. The blue laser beam passes through a collimator lens to become substantially parallel light, and then passes through a dichroic mirror to combine with the green laser beam. After the red laser beam passes through the collimator lens and becomes substantially parallel light, it is combined with the green laser beam and the blue laser beam through a dichroic mirror. The multiplexed laser light passes through the multi-lens to become diffused light, and enters the spatial modulation element. The spatial modulation element modulates the incident light based on the periodic main image signal. The projection lens projects the light modulated by the spatial modulation element on the ground, and gives the pattern color identification. In this embodiment, the visual device further includes an image correction controller, and the visual device captures the image displayed by the light projected by the projection lens. After the image correction controller is processed, it is converted into a machine-recognizable image or an image identifier that the robot has trained and learned in advance, and the processed projection pattern is sent to the robot in the nearby area.

In this embodiment, when the robot needs to work and starts to move, the robot sends an on message to the projection device. The opening message format is shown as follows: id, angel, time, shift. A message includes four fields, namely serial number, angle, time and switching bit. The serial number is used for identification, and the serial number of each message command is different, so as to distinguish different commands and control the corresponding robot and its projection device. , the angle is used to indicate the projection angle of the projection device, and the projection angle is changed according to the movement speed of the robot. The faster the robot runs, the farther the projected image is from the robot; time is used to control the duration of projection, such as when the robot travels in a spacious In the straight section, the projection time can be controlled intermittently; shift is the value of the switching bit, and the value of the switching bit attribute can be in many cases, that is, each value represents the pattern projected by the projection.

In this embodiment, when the robot's own vision device 70 captures the projection pattern projected by its own projection device 60 , that is, when the first robot 100 captures the first projection pattern carrying its own motion state, the first robot 100 You can do nothing by default, or check whether the projection pattern is displayed normally on its travel path. If it cannot capture its own projection pattern, it will give an abnormal feedback.

Based on the description of the embodiment shown in FIG. 6 , as shown in FIG. 7 , in another embodiment of the present invention, the robot further includes a processing device 80 for recognizing the projection pattern and controlling the robot to perform corresponding obstacle avoidance measures.

In this embodiment, a processing device 80 can be separately provided on the robot body to recognize the first projection pattern and perform further processing, such as recognizing projection pattern information, acquiring robot motion information corresponding to the projection pattern, and combining with its own robot The motion state is used for obstacle avoidance processing. In a possible embodiment, the processing device 80 only needs to be installed on the second robot body that needs to perform obstacle avoidance measures. According to the working space of the robot and the frequency of the road sections traveled by the robot, the path traveled by the robot is divided into main roads and secondary roads. The main roads are the roads that robots frequently pass through or perform important tasks. The road section or the road section where the secondary task is performed, the robot on the main road is the first robot 100, which can only project the projected pattern without capturing the projected pattern, and the robot on the secondary road is the second robot 200, which only captures the projected pattern without projecting the projected pattern, In other words, the first robot 100 drives on the main road, the second robot 200 drives on the secondary road, the second robot 200 actively avoids obstacles, and the first robot 100 can keep its desired speed to complete more tasks .

In another possible embodiment, the processing device 80 may be integrated or embedded in the robot central processing system in the form of a processing module, for identifying the first projection pattern, and combining its own motion information to perform obstacle avoidance control. Specifically, the second robot recognizes the first projection pattern through the processing module, obtains the movement information of the first robot, and then learns the movement speed, movement direction, and position of the first robot from the first projection pattern, combined with its own movement speed, direction, determine the approximate position where it meets the first robot, and calculate the collision probability. If the collision probability is greater than the preset value, it will take obstacle avoidance measures, such as deceleration, acceleration, stop, steering and other actions.

8 and 9, when there is no projection device, the first robot 100 travels at a right angle relative to the second robot 200, and the second robot 200 cannot observe the body of the first robot 100 due to the wall, and also cannot predict the first robot 100. The next travel route of a robot 100. The obstacle avoidance action is not triggered until the first robot 100 and the second robot 200 move within a relatively short distance and their lidars scan the outline of each other. At this time, the avoidance effect is poor, and it is possible that the avoidance is not timely and a collision occurs.

When there is a projection device, the second robot 200 projects a two-dimensional code pattern containing the semantics of "Robot JX02, going straight, speed 1.5m/s, distance 6m" on the ground in front of the second robot 200 . Although the lidar of the first robot 100 has not yet scanned the outline of the second robot 200, the vision device of the first robot has captured the two-dimensional code of the second robot, and recognizes it according to the two-dimensional code "Robot JX02, is going straight, The speed is 1.5m/s and the distance is 6m”. The semantics is calculated, the decision is made, and the avoidance action is taken in advance. The avoidance effect in this embodiment is good, especially when the robot is running at a high speed, it can obtain other robots earlier. Dynamic information, get more reaction time to make corresponding obstacle avoidance actions.

Referring to FIG. 9 again, in an optional embodiment of the present application, according to the map information of the robot operation, the first robot 100 is a robot that travels laterally, and the first robot 100 is provided with a projection device that projects a two-dimensional code pattern on the travel path. , the second robot 200 is a robot that travels longitudinally, and the second robot 200 is provided with a vision device to capture the two-dimensional code pattern on the travel path. In this optional embodiment, only the second robot 200 needs to perform obstacle avoidance actions , in order to avoid the first robot 100, the first robot 100 does not do any evasive action, and is only responsible for projecting its own motion information on the traveling ground, that is, only transmitting information, and the second robot 200 only receives information and makes decisions.

As shown in FIG. 10 , the present application discloses a multi-robot dynamic obstacle avoidance system. The obstacle avoidance system includes:

Several robots, several robots move autonomously in the same working environment;

The projection module 600, in response to the autonomous movement of the robot, generates different projection patterns in the travel path of the robot, and the projection patterns carry the motion information of the robot;

a vision module 700, configured to capture the projected pattern in the travel path of the robot;

The processing module 800 is configured to identify the motion information of the robot corresponding to the projection pattern, and control the other robot to perform corresponding obstacle avoidance measures in combination with the motion information of the other robot.

In this embodiment, the projection module 600 performs real-time projection along with the movement of the robot, and corresponds to the current movement state of the robot. Specifically, the preset distances between the first projection pattern and the robot are different distances such as 2 meters, 5 meters, and 10 meters. It can be understood that the greater the movement speed of the robot, the longer the projection distance. The motion information of the robot includes the robot's own code, speed, driving direction, distance from the projection pattern, etc. These information can be represented by different projection patterns. The projection pattern is a pattern that the robot can recognize its semantics, such as QR (Quick Response) code II. dimensional code form.

Further, the projection module 600 can project different projection patterns, and each projection pattern corresponds to different robot motion information. Specifically, several preset motion states that the robot likes can be used as the basic projection pattern, for example, a speed value is taken from the four stages of low speed, medium and low speed, medium speed, and high speed, and it is matched with turning left, turning right, going straight, and stopping. After waiting for the normal turning, take several projection distance values such as 2 meters, 3 meters, 5 meters, 10 meters, etc., extract the values from the above attributes, and add the robot's own code to form the basic projection pattern. The distance and speed are integrated into a time attribute, that is, the time interval for the robot to keep its motion state unchanged and reach the first projection pattern, such as 3s, 5s, etc., that is to say, the robot will reach the first projection pattern generated at this moment after 3s. The above basic projection patterns are taught to the robot in advance, so that in the actual operation scene, the robot can quickly identify the projection pattern, and make corresponding planning measures to avoid obstacles.

In this embodiment, the vision module 700 is used to capture the projection pattern within the visual range. The vision module 700 can be arranged on the robot, and along with the movement of the robot, dynamically capture the projection pattern on the movement path of the robot, or it can be relatively fixedly arranged on the robot In the working environment, the projection pattern of some areas is monitored in real time. In this embodiment, the processing module 700 can be set separately to identify the first projection pattern and perform further processing. For example, after identifying the projection pattern information and performing unified computing processing on the robot in the area similar to the projection pattern, then the area robot for scheduling.

Preferably, the processing module 800 is provided on the robot body, or the processing module 800 is integrated or embedded in the robot central processing system, for recognizing the first projection pattern and further integrating the processing. Specifically, the second robot recognizes the first projection pattern through the processing module 800, obtains the motion information of the first robot, and then understands the movement speed, the movement direction, and the position from the first projection pattern of the first robot, combined with its own movement speed , direction, determine the approximate position where it meets the first robot, and calculate the collision probability. If the collision probability is greater than the preset value, take obstacle avoidance measures, such as deceleration, acceleration, stop, steering and other actions.

The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included in the protection of the present invention. within the range.

Industrial Applicability

The invention provides a robot motion information identification method, an obstacle avoidance method, an obstacle avoidance robot, and an obstacle avoidance system. On the current travel path of the robot, a first projection pattern carrying the motion information of the robot is generated, the first projection pattern is visually captured, and the motion information of the robot carried by the first projection pattern is recognized. Among them, the motion information of the robot is transmitted through the projection pattern, and the visual capture projection pattern is used for identification processing, and obstacle avoidance planning is further adopted, so that the robot can obtain the driving information of other robots on the route before the lidar recognizes other robots. Pre-evasion can improve safety and reduce the probability of corner collision on the one hand; on the other hand, it can improve the running speed and increase efficiency on the basis of safety. Therefore, it has industrial applicability.

Claims

A method for identifying motion information of a robot, comprising:

On the current travel path of the robot, a first projection pattern carrying the motion information of the robot is generated;

visually capturing the first projected pattern;

Identify the robot motion information carried by the first projection pattern.
The information identification method according to claim 1, wherein, on the current travel path of the robot, generating the first projection pattern carrying the motion information of the robot comprises:

The first projection pattern is projected on the ground of the first robot along its travel direction; or/and the first projection pattern is projected by an external projection device in the working environment of the first robot.
The information identification method of claim 1, wherein the visually capturing the first projection pattern comprises:

The second robot captures the first projection pattern through a vision device installed on its body; or/and a vision device installed in the robot working environment to capture the first projection pattern.
The information identification method according to claim 1, wherein the identifying the first robot motion information carried by the first projection pattern comprises:

The processing device provided on the robot receives the first projection pattern, and identifies the motion information of the first robot corresponding to the projection pattern; or/and

The processing device installed in the robot working environment receives the first projection pattern, and identifies motion information of the first robot corresponding to the first projection pattern.
A multi-robot motion obstacle avoidance method, comprising:

The first robot projects a first projection pattern carrying its own motion information on the ground along its travel direction;

The second robot captures the first projected pattern through a vision device installed on its body;

The second robot recognizes the motion information of the first robot corresponding to the first projection pattern, and performs corresponding obstacle avoidance measures in combination with its own motion information to avoid collision with the first robot.
The obstacle avoidance method according to claim 5, wherein the second robot performs corresponding obstacle avoidance measures in combination with its own motion information, comprising:

Based on the optimal mutual collision avoidance algorithm, the second robot performs obstacle avoidance planning according to the current position and speed of the first robot or/and the position and speed of the first robot in the first projection pattern, and changes own speed, direction; or

The second robot calculates the first time when the first robot reaches the first projection pattern, and changes its movement speed according to a preset time interval to make it advance or delay the first time by the time interval The first projection pattern is reached.
The obstacle avoidance method according to claim 5, wherein after the second robot performs corresponding obstacle avoidance measures in combination with its own motion information, it further comprises:

The second robot projects a second projection pattern on the ground along its travel direction after the obstacle avoidance measure.
A robot capable of dynamic obstacle avoidance, comprising:

The main body of the robot can be moved autonomously;

a projection device, which is relatively fixedly arranged on the robot body; the projection device includes an electronic display and a plurality of projection films for the robot to project different projection patterns;

A vision device, which is installed on the robot body on the same side as the projection device, and is used to capture the projection pattern of the robot's travel area.
The robot according to claim 8, wherein the robot further comprises a processing device for recognizing the projection pattern and controlling the robot to perform corresponding obstacle avoidance measures.
A multi-robot obstacle avoidance system, comprising: a plurality of robots that move autonomously in the same working environment;

a projection module, in response to the autonomous motion of the robot, generating different projection patterns in the travel path of the robot, the projection patterns carrying the motion information of the robot;

a vision module for capturing the projected pattern in the travel path of the robot;

The processing module is used for identifying the motion information of the robot corresponding to the projection pattern, and combining the motion information of the other robot to control the other robot to perform corresponding obstacle avoidance measures.