Mobile mechanical arm collision detection method based on point cloud data

Abstract

The invention provides a mobile mechanical arm collision detection method based on point cloud data. The movable mechanical arm is composed of a movable chassis and an airborne mechanical arm, and the movable chassis carries a radar and an IMU sensor to obtain working site environment information and the posture of the movable chassis. And point cloud data is stored when entering an operation stage, is converted to a base coordinate system of the airborne mechanical arm, and is taken as an initial frame after downsampling and filtering processing, so that the performance of subsequent collision detection is improved. Information of the odometer is obtained in real time, and dynamic updating of point cloud data is achieved through transformation between the current position and the initial frame position. And the point cloud data updated in real time is converted into an Octomap, and whether the planned action of the airborne mechanical arm will collide with an obstacle in the current environment map or not is continuously judged.

Description

Technical Field

[0001] The invention belongs to the technical field of robots, and in particular relates to a mobile robot arm collision detection method based on point cloud data. Background Art

[0002] With the rapid development of artificial intelligence, machine learning, sensor technology and robotics technology, the application of mobile robotic arms is becoming more and more extensive. Mobile robots are usually equipped with airborne robotic arms. In addition to realizing functions such as path planning, autonomous navigation, and environmental perception, they can also complete various operations such as handling, assembly, and welding, thereby improving operational efficiency and safety. They are mainly used in manufacturing, logistics, services, and agriculture.

[0003] The collision detection of the mobile robot arm during operation can effectively prevent the mobile robot arm from colliding with surrounding obstacles, ensure the continuous and efficient operation of the mobile robot arm, and protect the safety of personnel and equipment. Therefore, collision detection is a key technology and important guarantee for improving operation efficiency and adapting to environmental changes.

[0004] The current collision detection method for mobile robot arms mainly relies on the sensor data processing of the mobile chassis, such as lidar, ultrasonic sensors and cameras. However, this method usually faces the following important problems:

[0005] 1. The environmental information obtained by the mobile chassis sensor cannot reflect the actual working environment of the robot arm;

[0006] 2. The environmental information perceived by traditional methods is too redundant and complex to process;

[0007] 3. There is a delay in data processing and insufficient real-time performance, which reduces the reliability of collision detection.

[0008] Therefore, it is unreliable for mobile robotic arms to rely solely on traditional sensor data for collision detection. Other methods are needed to update the working environment information in real time to achieve collision detection. Summary of the invention

[0009] In view of the above problems, the present invention proposes a mobile manipulator collision detection method based on point cloud data. The mobile manipulator consists of a mobile chassis and an airborne manipulator, wherein the mobile chassis is equipped with radar and IMU sensors to obtain the working site environment information and the posture of the mobile chassis. When entering the working stage, the point cloud data is saved, converted to the base coordinate system of the airborne manipulator, and used as the initial frame after downsampling and filtering. The odometer information is obtained in real time, and the dynamic update of the point cloud data is realized by the transformation between the current position and the initial frame position. The point cloud data updated in real time is converted into an Octomap map, and it is continuously judged whether the planned action of the airborne manipulator will collide with the obstacles in the current environment map. The present invention reflects the real working environment information of the mobile manipulator by converting the environmental data collected by the mobile chassis to the base coordinate system of the airborne manipulator; by downsampling and filtering the point cloud data, redundant information is reduced and the performance of subsequent collision detection is improved; the point cloud data is updated by a simple coordinate conversion between the current position and the position when the initial frame is collected, the data processing efficiency is improved, and the real-time performance is guaranteed.

[0010] The technical solution adopted by the present invention to solve its technical problem is: a mobile robot arm collision detection method based on point cloud data, characterized in that it includes the following steps: Step 1: The mobile chassis is equipped with a radar to perceive the environment of the work site, and is equipped with an IMU to obtain the posture of the mobile chassis in real time; Step 2: Calibrate to obtain the transformation matrix between the radar and the IMU and between the IMU and the center of the front wheel axle of the mobile chassis, and measure to obtain the transformation matrix between the center of the front wheel axle of the mobile chassis and the base coordinate system of the robot arm; Step 3: When entering the operation stage, save the point cloud data of the current frame, and record the odometer information of the front wheel of the chassis at that time, and convert the point cloud data of the current frame to the base coordinate system of the robot arm through coordinate transformation as the initial The first frame is captured, and the collected point cloud data is downsampled and filtered; Step 4: Real-time acquisition of the odometer information of the front wheel of the mobile robot, calculation of the transformation matrix between the current position and the position at the time of initial frame acquisition, and real-time update of the point cloud data is achieved through coordinate transformation; Step 5: Convert the point cloud data updated in real time into Octomap to generate a real-time updated environment map; Step 6: Real-time reception of the operation planning angle of the onboard robotic arm, continuous judgment on whether the current planned action will collide with obstacles in the current environment map, if there is no collision, the planning angle is issued, and the mobile robotic arm executes the planned action; otherwise, the current action is stopped to prevent collision, and a re-planning action signal is returned.

[0011] Step one includes: the research object of the present invention is the collision detection problem during the operation of a mobile robotic arm. The mobile robotic arm is composed of a mobile chassis and an airborne robotic arm, wherein the mobile chassis is equipped with a radar and an IMU. The radar provides point cloud data of the surrounding environment, and the IMU provides the posture of the mobile robot.

[0012] Step 2 includes: calibrating the transformation matrix T between the radar and the IMU and between the IMU and the center of the front wheel axle of the mobile chassis lti and T itr , measure and obtain the transformation matrix T between the front wheel axle center of the mobile chassis and the base coordinate system of the robotic arm rtb .

[0013] Step 3 includes: when entering the operation phase, save the point cloud data P of the current frame b , record the odometer position information of the current vehicle front wheel (t xini ,t yini , t zini ,) and posture information q ini =(w, x, y, z), and converted into matrix form

[0014]

[0015] Convert the point cloud data of the current frame to the robot base coordinate system through coordinate transformation

[0016] P ini =P b ·T lti ·T itr ·T rtb

[0017] This frame of point cloud will be used as the initial frame to provide reference for subsequent real-time position updates. The collected point cloud data is downsampled and filtered to ensure that the point cloud data is clean and reduce unnecessary redundancy, thereby improving the performance of subsequent collision detection.

[0018] Step 4 includes: obtaining the odometer information of the front wheel of the mobile chassis in real time and converting it into a matrix T real , calculate the transformation matrix between the current position and the position at the time of initial frame acquisition

[0019] T real to ini =T real -1 ·T ini

[0020] Real-time update of point cloud data through coordinate transformation

[0021] P real (t) = P ini ·T real to ini

[0022] Step five includes: converting the point cloud data updated in real time into Octomap to generate a real-time updated environment map.

[0023] Step six includes: receiving the operation planning angle of the airborne robotic arm in real time, and continuously judging whether the current planned action will collide with obstacles in the current environment map.

[0024] The beneficial effect of the present invention is that a mobile robotic arm collision detection method based on point cloud data is proposed, which solves the problems of limited perception accuracy of mobile chassis sensors, inaccurate environmental information, redundant point cloud data, poor real-time performance, low reliability and other problems in the traditional collision detection method during the autonomous operation of the mobile robotic arm, and has certain promotion significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of the present invention;

[0026] Figure 2 The robot platform of the present invention;

[0027] Figure 3 It is the point cloud information of the environment;

[0028] Figure 4 Octomap map converted from point cloud data that is updated in real time after processing; DETAILED DESCRIPTION

[0029] Reference Figure 1 A mobile robot arm collision detection method based on point cloud data is characterized in that it includes the following steps: Step 1: The mobile chassis is equipped with a radar to perceive the environment of the work site, and is equipped with an IMU to obtain the posture of the mobile chassis in real time; Step 2: Calibrate the transformation matrix between the radar and the IMU and the IMU and the center of the front wheel axle of the mobile chassis, and measure the transformation matrix between the center of the front wheel axle of the mobile chassis and the base coordinate system of the robot arm; Step 3: When entering the operation stage, save the point cloud data of the current frame, and record the odometer information of the front wheel of the chassis at that time, convert the point cloud data of the current frame to the base coordinate system of the robot arm as the initial frame through coordinate transformation, and collect The point cloud data is downsampled and filtered; Step 4: Obtain the odometer information of the front wheel of the mobile robot in real time, calculate the transformation matrix between the current position and the position at the time of initial frame acquisition, and realize the real-time update of the point cloud data through coordinate transformation; Step 5: Convert the point cloud data updated in real time into Octomap to generate a real-time updated environment map; Step 6: Receive the operation planning angle of the onboard robotic arm in real time, and continuously judge whether the current planned action will collide with the obstacles in the current environment map. If there is no collision, send the planning angle and move the robotic arm to execute the planned action; otherwise, stop the current action to prevent collision and return the re-planning action signal.

[0030] Step one includes: the research object of the present invention is the collision detection problem during the operation of a mobile robotic arm. The mobile robotic arm is composed of a mobile chassis and an airborne robotic arm, wherein the mobile chassis is equipped with a radar and an IMU. The radar provides point cloud data of the surrounding environment, and the IMU provides the posture of the mobile robot.

[0031] Step 2 includes: calibrating the transformation matrix T between the radar and the IMU and between the IMU and the center of the front wheel axle of the mobile chassis lti and T itr , measure and obtain the transformation matrix T between the front wheel axle center of the mobile chassis and the base coordinate system of the robotic arm itb The transformation relationship between multiple coordinate systems is established in order to unify the point cloud data and the mobile chassis posture information provided by the IMU into the target coordinate system.

[0032] Step 3 includes: when entering the operation phase, save the point cloud data P of the current frame b , record the odometer position information of the current chassis front wheel (t xini , t yini , t zini ) and posture information q ini =(w, x, y, z) and convert it into matrix form

[0033]

[0034] Convert the point cloud data of the current frame to the robot base coordinate system through coordinate transformation

[0035] P ini =P b ·T lti ·T itr ·T rtb

[0036] This frame of point cloud will be used as the initial frame to provide reference for subsequent real-time position updates. The collected point cloud data is downsampled and filtered to ensure that the point cloud data is clean and reduce unnecessary redundancy, thereby improving the performance of subsequent collision detection.

[0037] Step 4 includes: obtaining the odometer information of the front wheel of the mobile chassis in real time and converting it into a matrix T real , calculate the transformation matrix between the current position and the position at the time of initial frame acquisition

[0038] T real to ini =T real -1 ·T ini

[0039] Real-time update of point cloud data is achieved through coordinate transformation, ensuring that the chassis’ position information is synchronized with environmental data.

[0040] Preal (t) = P ini ·T real to ini

[0041] Step 5 includes: converting the real-time updated point cloud data into Octomap to generate a real-time updated environment map

[0042] O = Octomap (P real (t))

[0043] Octomap can effectively record and update obstacles in the environment, facilitating real-time updates and collision detection in dynamic environments.

[0044] Step 6 includes: receiving the operation planning angle of the onboard manipulator in real time, and continuously judging whether the current planned action will collide with obstacles in the current environment map

[0045]

[0046] If there is no collision The planned angle is sent down, and the robot arm executes the planned action; otherwise, the current action is stopped and re-planned. Dynamic collision detection and control of mobile robot operations are realized to ensure operation safety.

[0047] The above is only a specific embodiment of the present invention, and cannot be used to limit the scope of the invention. Therefore, the replacement of equivalent components, or equivalent changes and modifications made according to the scope of protection of the patent of the present invention, should still fall within the scope of this patent. In addition, the technical features of the present invention can be freely combined with each other, with each other, and with each other.

Claims

1. A mobile robot arm collision detection method based on point cloud data, characterized in that: The following steps are included: Step 1: The mobile chassis is equipped with a radar to perceive the environment at the work site, and is equipped with an IMU to obtain the posture of the mobile chassis in real time; Step 2: Calibrate the transformation matrix between the radar and the IMU and between the IMU and the center of the front wheel axle of the mobile chassis, and measure the transformation matrix between the center of the front wheel axle of the mobile chassis and the base coordinate system of the robotic arm; Step 3: When entering the operation stage, save the point cloud data of the current frame, and record the odometer information of the front wheel of the chassis at that time. Through coordinate transformation, the point cloud data of the current frame is converted to the base coordinate system of the robotic arm as the initial frame, and the collected point cloud data is downsampled and filtered. ; Step 4: Obtain the odometer information of the front wheel of the mobile robot in real time, calculate the transformation matrix between the current position and the position at the time of initial frame acquisition, and realize real-time update of point cloud data through coordinate transformation; Step 5: Convert the real-time updated point cloud data into Octomap to generate a real-time updated environment map; Step 6: Receive the operation planning angle of the onboard robotic arm in real time, and continuously judge whether the current planned action will collide with the obstacles in the current environment map. If there is no collision, send the planning angle and move the robotic arm to execute the planned action; otherwise, stop the current action to prevent collision and return a re-planning action signal.

2. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step one includes: the research object of the present invention is the collision detection problem during the operation of a mobile robotic arm. The mobile robotic arm is composed of a mobile chassis and an airborne robotic arm, wherein the mobile chassis is equipped with a radar and an IMU. The radar provides point cloud data of the surrounding environment, and the IMU provides the posture of the mobile robot.

3. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step 2 includes: calibrating the transformation matrix T between the radar and the IMU and between the IMU and the center of the front wheel axle of the mobile chassis lti and T itr , measure and obtain the transformation matrix T between the front wheel axle center of the mobile chassis and the base coordinate system of the robotic arm rtb , the transformation relationship between multiple coordinate systems is established, and the point cloud data and the mobile chassis posture information provided by IMU are unified into the target coordinate system.

4. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step 3 includes: when entering the operation phase, save the point cloud data P of the current frame b , record the odometer position information of the current chassis front wheel (t xini , t yini , t zini ,) and posture information q ini =(w, x, y, z), and converted into matrix form Convert the point cloud data of the current frame to the robot base coordinate system through coordinate transformation P ini =P b ·T lti ·T itr ·T rtb This frame of point cloud will serve as the initial frame to provide a reference for subsequent real-time position updates. The collected point cloud data will be downsampled and filtered to ensure that the point cloud data is clean and reduce unnecessary redundancy, thereby improving the performance of subsequent collision detection.

5. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step 4 includes: obtaining the odometer information of the front wheel of the mobile chassis in real time and converting it into a matrix T real , calculate the transformation matrix between the current position and the position at the time of initial frame acquisition T real to ini =T real -1 ·T ini Real-time update of point cloud data is achieved through coordinate transformation, ensuring that the chassis posture information is synchronized with the environmental data P real (t)=P ini ·T real to ini。 6. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step 5 includes: converting the real-time updated point cloud data into Octomap to generate a real-time updated environment map O=Octomap(P real (t)) Octomap can effectively record and update obstacles in the environment, facilitating real-time updates and collision detection in dynamic environments.

7. The mobile robot arm collision detection method based on point cloud data according to claim 1, characterized in that: Step 6 includes: receiving the operation planning angle of the onboard manipulator in real time, and continuously judging whether the current planned action will collide with obstacles in the current environment map If there is no collision The planned angle is issued, and the robot arm executes the planned action; otherwise, the current action is stopped and re-planned, realizing dynamic collision detection and control of the mobile robot arm operation to ensure operation safety.

Images