CN-121798638-B - Self-adaptive environment-aware rail transit intelligent inspection robot navigation system

CN121798638BCN 121798638 BCN121798638 BCN 121798638BCN-121798638-B

Abstract

The invention belongs to the technical field of inspection robots, and particularly relates to a self-adaptive environment-aware rail transit intelligent inspection robot navigation system, which comprises a joint state acquisition module, a contact characteristic extraction module, a navigation factor calculation module, an impedance adjustment and chassis deviation correction module, wherein the joint state acquisition module is used for acquiring the joint state of the rail transit intelligent inspection robot; the method comprises the steps of obtaining a joint state vector and an electromagnetic driving moment vector, calculating external contact moment according to the joint state vector and the electromagnetic driving moment vector, obtaining a contact force characteristic index of the tail end of the mechanical arm, obtaining a transverse deviation navigation factor of the position and the posture of the chassis according to the difference between the contact force characteristic index and a reference force index, obtaining the self-adaptive deviation correction angular velocity of the movable chassis according to the transverse deviation navigation factor, and obtaining a self-adaptive target stiffness value according to the time domain fluctuation characteristic of the contact force characteristic index for navigation. The navigation method and the navigation device improve navigation precision and environmental robustness of the inspection robot.

Inventors

SHI LINGFENG
CAO DAGUANG
WU YUPING
WU GAO
CHEN ZHUOLIN
LUO GUANGXUN

Assignees

广东华能机电集团有限公司

Dates

Publication Date: 20260508
Application Date: 20260309

Claims (8)

1. Self-adaptation environment-aware track traffic intelligent inspection robot navigation system, its characterized in that includes: the joint state acquisition module acquires a joint state vector and an electromagnetic driving moment vector of the mechanical arm through a sensor at the joint of the mechanical arm; the acquisition mode of the electromagnetic driving moment vector comprises the steps of reading real-time torque current components through motor drivers of all joints to construct a real-time torque current vector, acquiring preset torque constants of all joint motors of the mechanical arm, constructing a diagonal matrix of the motor torque constants according to the preset torque constants, constructing a reduction ratio matrix according to specification parameters of all joint speed reducers, and mapping the real-time torque current vector into the electromagnetic driving moment vector by combining preset electromechanical conversion efficiency coefficients, the reduction ratio matrix and the diagonal matrix of the motor torque constants; The contact characteristic extraction module inputs the joint state vector into the dynamic model to obtain an internal dynamic moment, and comprises the steps of taking the joint state vector as input, calculating to obtain an inertial force item vector, a Coriolis force and centrifugal force item vector and a theoretical gravity item vector of the mechanical arm under the current gesture by using a recursion Newton-Euler algorithm, constructing a preset coulomb-viscous friction model, calculating to obtain an internal friction moment vector of each joint of the mechanical arm based on an angular velocity component in the joint state vector and a preset friction coefficient matrix, summing the theoretical gravity item vector, the Coriolis force and centrifugal force item vector, the inertial force item vector and the internal friction moment vector, and obtaining an internal dynamic moment for counteracting the stress of the mechanical arm; Stripping the internal dynamic moment from the electromagnetic driving moment vector to obtain an external contact moment, and acquiring a contact force characteristic index based on the external contact moment; The navigation factor calculation module is used for obtaining the difference between the contact force characteristic index and a preset reference force index, and carrying out smoothing treatment on the difference by utilizing an index weighted moving average model to obtain a transverse deviation navigation factor of the chassis pose; The impedance adjustment and chassis deviation correction module obtains self-adaptive deviation correction angular velocity according to the transverse deviation navigation factor, obtains self-adaptive target stiffness value of the mechanical arm joint according to time domain fluctuation characteristics of the contact force characteristic index, and drives the mobile chassis to correct the pose by combining the self-adaptive deviation correction angular velocity.
2. The adaptive environment-aware rail transit intelligent inspection robot navigation system according to claim 1, wherein the acquiring of the joint state vector of the mechanical arm by the sensor at the joint of the mechanical arm comprises controlling the mechanical arm to keep a preset flexible probe gesture so that a roller assembly at the tail end of the mechanical arm keeps contact with the side wall of the rail, and acquiring real-time rotation angles and angular velocities of the joints by the absolute value encoder installed at the joints of the mechanical arm to construct the joint state vector.
3. The adaptive environment-aware rail transit intelligent patrol robot navigation system according to claim 1, wherein the step of stripping the internal dynamic moment from the electromagnetic driving moment vector to obtain the external contact moment comprises calculating the external contact moment acting on the joint space of the mechanical arm according to the joint state vector, the electromagnetic driving moment vector, the pre-stored Lagrangian dynamic parameters of the robot and the friction coefficient through dynamic compensation based on a recursive Newton-Euler algorithm and filtering observation of a momentum observer.
4. The self-adaptive environment-aware rail transit intelligent patrol robot navigation system according to claim 1 is characterized in that the acquisition mode of the contact force characteristic index comprises the steps of mapping external contact moment of a mechanical arm joint space to a Cartesian space by using pseudo inverse of a mechanical arm jacobian matrix transpose to obtain a terminal contact force vector, performing modular length calculation on the terminal contact force vector, and acquiring the contact force characteristic index according to the modular length of the terminal contact force vector.
5. The adaptive environment-aware rail transit intelligent patrol robot navigation system according to claim 4, wherein the method for obtaining the lateral deviation navigation factor comprises the steps of performing differential calculation on a contact force characteristic index and a reference force index to obtain an original force sense residual, performing iterative smoothing on the original force sense residual by using an index weighted moving average model to obtain a smooth force sense residual, and obtaining the lateral deviation navigation factor according to the smooth force sense residual.
6. The adaptive environment-aware rail transit intelligent inspection robot navigation system according to claim 5, wherein the obtaining the adaptive deviation correcting angular velocity according to the lateral deviation navigation factor comprises determining a deviation correcting rotation direction according to the positive and negative of the lateral deviation navigation factor, and obtaining the adaptive deviation correcting angular velocity by combining a preset basic deviation correcting angular velocity gain, the deviation correcting rotation direction and the lateral deviation navigation factor.
7. The self-adaptive environment-aware rail transit intelligent inspection robot navigation system according to claim 1 is characterized in that the self-adaptive target stiffness value of the mechanical arm joint is obtained according to time domain fluctuation characteristics of contact force characteristic indexes, the self-adaptive target stiffness value comprises the steps of collecting and storing a contact force characteristic index sequence in a time sliding window in real time, calculating variance of the contact force characteristic index sequence, constructing a target stiffness correction coefficient by utilizing the variance, and adjusting a preset basic stiffness value by utilizing the target stiffness correction coefficient to obtain the self-adaptive target stiffness value.
8. The intelligent track traffic inspection robot navigation system based on the self-adaptive environment sensing is characterized in that the intelligent track traffic inspection robot navigation system based on the self-adaptive environment sensing is used for driving a mobile chassis to conduct pose correction according to the self-adaptive correction angular speed, comprises the steps of updating stiffness matrix parameters in an impedance control law in real time according to self-adaptive target stiffness values, smoothing high-frequency impact generated by environment contact through physical flexibility of a mechanical arm, driving the mobile chassis to turn based on the self-adaptive correction angular speed, and guiding the robot to travel along a track center line while keeping the mechanical arm in flexible contact with a tunnel side wall.

Description

Self-adaptive environment-aware rail transit intelligent inspection robot navigation system Technical Field The invention relates to the technical field of inspection robots. More particularly, the invention relates to a self-adaptive environment-aware rail transit intelligent patrol robot navigation system. Background Along with the rapid expansion of the track traffic network, in order to ensure the safe operation of equipment and lines in a tunnel, track traffic tunnel inspection robots are often applied to daily inspection maintenance work. In the inspection process, the robot needs to keep a stable running track along the side wall of the track or tunnel so as to ensure that the detection equipment can effectively cover the target area. Therefore, the navigation system with high precision and high reliability is a core guarantee for the inspection robot to complete the task. In the related art, for example, a chinese patent document with a bulletin number CN109176513B discloses a method and a system for inspecting an intelligent inspection robot, which includes training a specific part of a train by an identification algorithm to obtain feature data and set a preset image, acquiring an image of the bottom of the train in real time during inspection, matching and comparing the feature point with the preset image, thereby identifying the specific part and obtaining an inspection result, and controlling the robot to travel to the next feature part according to distance information between adjacent feature parts by using a position tracking unit, so as to realize an automatic inspection process. However, the related art has significant limitations in a complex track traffic tunnel environment. Firstly, tunnel environment is extremely bad, high-concentration metal dust, water mist and violently-changed illumination conditions exist throughout the year, and the factors can seriously interfere with the imaging quality of a vision sensor, so that the image characteristic points are difficult to extract or are wrong to match, and therefore, a navigation method based on visual image comparison is invalid, and the positioning precision of a robot cannot be ensured. In the prior art, the physical contact state between the robot and the side wall of the tunnel cannot be directly obtained mainly by non-contact visual perception, when the robot is required to keep a specific gesture close to the side wall for fine operation, a visual system is difficult to perceive small transverse deviation caused by uneven track or slipping of a chassis in real time, the robot and the wall surface of the tunnel are easy to generate rigid collision or unexpected detachment, and the adaptability to physical interaction of the environment is lacking. Disclosure of Invention In order to solve the technical problems of low positioning precision, lack of environment interaction sensing capability and easy occurrence of rigid collision or derailment caused by failure of vision and inertial navigation sensors in tunnel environments with high dust, water mist and strong electromagnetic interference of the track inspection robot in the prior art, the invention provides a self-adaptive environment sensing track traffic intelligent inspection robot navigation system, which comprises a joint state acquisition module, a control module and a control module, wherein the joint state acquisition module acquires a joint state vector and an electromagnetic driving moment vector of a mechanical arm through a sensor at the joint of the mechanical arm; the system comprises a contact characteristic extraction module, a navigation factor calculation module, an impedance adjustment and chassis correction module, a mechanical arm joint self-adaption target stiffness value and a mobile chassis driving gesture correction combining self-adaption angular velocity, wherein the contact characteristic extraction module inputs a joint state vector into a dynamic model to obtain an internal dynamic moment, strips the internal dynamic moment from an electromagnetic driving moment vector to obtain an external contact moment, and obtains a contact force characteristic index based on the external contact moment, the navigation factor calculation module obtains the difference between the contact force characteristic index and a preset reference force index, and performs smoothing treatment on the difference by utilizing an index weighted moving average model to obtain a transverse deviation navigation factor of the chassis gesture, the impedance adjustment and chassis correction module obtains self-adaption correction angular velocity according to the transverse deviation navigation factor, and obtains the self-adaption target stiffness value of the mechanical arm joint according to the time domain fluctuation characteristic of the contact force characteristic index. According to the invention, the external contact moment is inverted by collecting the internal current sig