Robot capable of keeping bearing surface horizontal when going up and down stairs and control method of robot
Through the alternate structure of wheel-shops and the coordinated control of the depth camera, the inclination problem of climbing robots when climbing and going down the stairs is solved, the load-bearing surface level is maintained, stability and efficiency are improved, and adaptability and autonomy are enhanced.
Patent Information
- Application Number
- CN202411095512.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-19
AI Technical Summary
Existing stair climbing robots are prone to shifting the center of gravity when going up and down stairs, resulting in stability problems, and slow climbing speed, low adaptability and efficiency.
Adopting a unique alternate structure of wheel-tracking, combining mechanical legs and track chassis, through the collaborative control of the front-end and rear-end depth cameras, the robot maintains the level of the load-bearing surface when going up and down the stairs, and adjusts the robot's posture with a telescopic rod, combining deep learning and computers. Visually identify stair information and accurately plan the path.
It significantly improves the stability and safety of the robot during stairs climbing, ensures the smooth progress of inspection tasks, improves autonomy and work efficiency, and enhances the ability to adapt to different terrains.
Smart Images

Figure CN118977778A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of inspection robots, and in particular relates to a robot capable of keeping the load-bearing surface of stairs going up and down horizontally and a control method thereof. Background Art
[0002] Indoor inspection robot technology, as an important branch of the robotics field, specializes in inspection, monitoring and safety protection in indoor environments. It is particularly good at performing inspection tasks in complex terrains such as going up and down stairs, ensuring all-round safety management without blind spots.
[0003] In the prior art, the Chinese invention patent with application number CN202211490008.1 discloses an indoor wheeled inspection robot, which can turn flexibly and run stably on flat ground, but cannot complete cross-floor inspection tasks in multi-story buildings. Invention patent CN102477268A discloses an under-actuated tracked mobile obstacle-crossing platform, whose track has poor maneuverability and has defects when moving on the ground. CN111232081A discloses an omnidirectional horizontal posture climbing robot, which climbs the stairs alternately by a forward unit group and a vertical motion unit group, but the climbing speed is slow and the obstacle-crossing ability is low. CN109178137A discloses a multi-legged robot based on a three-degree-of-freedom mechanical leg. When climbing the stairs, the motor drives the mechanical legs to climb step by step, the control is complex, and the speed is slow. CN109927808A discloses a cooperative tracked quadruped robot dog, which has weak adaptability. Although the structure is simple, the climbing speed is slow. In addition, the Chinese invention patent with application number CN202110121084.4 can achieve stair climbing, but the robot body is tilted when going upstairs, which easily causes the center of gravity to shift. This design has stability issues when carrying goods or equipment.
[0004] At present, a lot of results have been achieved in the theoretical research and product development of mobile stair-climbing robot platforms, and some products have been launched on the market. However, most existing stair-climbing robots have problems such as poor carrying capacity and slow climbing speed. Although the quadruped obstacle-climbing robot is flexible, it is complex to control and not easy to popularize. Although the tracked robot has good stability, its mobility is poor. In response to this problem, the present application proposes a new robot design. The robot adopts a unique wheel-track alternating structure, which significantly enhances its flexibility, enabling it to easily go up and down stairs and maintain the horizontal stability of the load-bearing surface. At the same time, the robot can also move smoothly within the floor, thereby effectively performing various cross-floor tasks. Summary of the invention
[0005] The purpose of the present invention is to solve the problem of main body tilt and center of gravity shift of the current stair climbing robot when going up and down stairs, and proposes a robot with horizontal load-bearing surface when going up and down stairs and a control method thereof.
[0006] In order to achieve the above-mentioned object, in a first aspect, the present invention provides a robot for keeping the load-bearing surface of stairs horizontal, which adopts the following technical solution:
[0007] A robot for keeping the load-bearing surface level when going up and down stairs, comprising a robot shell and a front-end depth camera and a rear-end depth camera respectively fixed to the front and rear ends of the robot shell;
[0008] There are two mechanical legs on each side of the robot housing, the top of each mechanical leg is equipped with a motor to control its rotation, and the bottom of each mechanical leg is equipped with a wheel motor connected to the wheel through a flange;
[0009] The robot has a crawler chassis at the bottom of the shell, and a rear wheel fixing bracket is installed at the rear. Two small auxiliary wheels are installed on both sides of the bracket through bearings for use when walking on flat ground. In addition, the crawler chassis is connected to the telescopic rod through a hinge, and the telescopic rod is also connected to the robot shell through a hinge. During the stair climbing movement, the angle of the robot's load-bearing surface can be adjusted through the telescopic rod to keep it horizontal;
[0010] The crawler chassis adopts a symmetrical design and consists of two drive motors, two crawlers and a bottom plate. Each drive motor is responsible for driving a driving wheel. Each crawler is equipped with three load-bearing wheels, which are located on both sides and fixedly connected to the bottom plate through their three support shafts. In addition, a guard plate is fixed on the outside of each load-bearing wheel to provide protection;
[0011] Furthermore, the mechanical legs of the robot are directly driven by motors, and the wheels at the bottom of the mechanical legs are independently driven by wheel motors to control their rotation and adjust the robot's posture;
[0012] Furthermore, the two tracks of the track chassis are respectively driven by a drive motor, and the drive motor controls the rotation speed and direction of the tracks, so that the robot can smoothly go up and down stairs and cross obstacles;
[0013] Furthermore, the front-end depth camera and the rear-end depth camera are respectively fixed to the front and rear ends of the robot housing, and can detect environmental information and stair data. The data is used by the control method proposed in this paper, so that the robot can autonomously plan the path, stably go up and down the stairs to cross the floors, and achieve the inspection task;
[0014] Furthermore, the control system of the robot realizes coordinated control of the robot's mechanical legs, tracked chassis and small auxiliary wheels by integrating the data of the front-end depth camera and the sensor information, thereby ensuring the robot's stable walking and floor crossing in different environments.
[0015] The robot, which can maintain the horizontal load-bearing surface when going up and down stairs, has a compact structure, can move flexibly in complex environments, is suitable for a variety of inspection scenarios, and has high autonomy and adaptability.
[0016] In order to achieve the above-mentioned purpose, in a second aspect, the present invention also provides a robot control method for keeping the load-bearing surface of stairs level, which adopts the following technical solution:
[0017] A robot control method for keeping the load-bearing surface of stairs going up and down horizontally adopts a robot for keeping the load-bearing surface of stairs going up and down horizontally as described in the first aspect. First, the robot collects image data through a front-end depth camera and a rear-end depth camera, and records the image and depth information of the current frame; then, the system scans each pixel in the image to detect whether there is a missing or blurred area of depth information; if all pixels have valid depth values and the image clarity meets the preset standard, the collection is stopped; otherwise, wait for the next frame of the image, use the optical flow algorithm to track the pixels, map the depth information of the next frame to the hole points of the previous frame to supplement the missing information, and ensure that the time interval between frames is short to maintain motion continuity; this process is repeated until all the hole points are filled and the image clarity meets the standard, and finally the noise and outliers are removed by filter processing, and the depth map is smoothed to obtain accurate and stable depth information;
[0018] Furthermore, we collected and annotated image datasets of stairs going up and down, used the dataset to train the YOLO V8 network, adjusted the algorithm's built-in parameters and network architecture to improve recognition accuracy, and used the trained YOLO V8 network model to process the images acquired by the front-end depth camera to identify the location and category of the stairs in the image. The specific process includes: identifying the edge points of the stairs in the image, converting the two-dimensional pixel coordinates into three-dimensional real coordinates through the internal and external parameters of the front-end depth camera, and constructing a three-dimensional model of the stairs;
[0019] Furthermore, after the three-dimensional data is processed, key point data of the stairs are extracted, and the edges of the stairs are identified using algorithms such as Canny edge detection; the edge of each step of the stairs is identified using the Hough transform algorithm; the number of parallel lines identified by the Hough transform is counted, which is the number of steps of the stairs;
[0020] Furthermore, the robot uses the stair information to perform stair climbing motion, and the specific steps include:
[0021] Step 1: When the robot moves to the front of the stairs, the mechanical legs rotate backwards, and the wheels rotate to push the robot forward. As the mechanical legs rotate, the track chassis contacts the stair surface and changes to a stair climbing posture. The robot shell is supported by a telescopic rod to keep it horizontal. The specific rotation angle θ(t) of the mechanical legs changes with time t as follows:
[0022] θ(t)=θ0+ω1t;
[0023] Where θ0 is the initial angle of the robotic leg, ω1 is the rotation speed of the robotic leg;
[0024] Step 2: When the mechanical leg rotates to a position parallel to the crawler chassis, the mechanical leg stops rotating to prevent the leg from getting stuck on the stairs. At this time, the rotation angle θ satisfies:
[0025] θ = 90°;
[0026] Step 3: The track starts to rotate under the drive of the driving wheel, pushing the robot to climb the stairs step by step. The relationship between the linear velocity v of the track and the rotational velocity ω2 is:
[0027] V = r·ω2;
[0028] Where r is the radius of the driving wheel;
[0029] Step 4: When the robot is about to reach the last two steps, the mechanical legs start to rotate again and gradually return to the original position. At the same time, the length of the telescopic rod is adjusted to keep the robot shell horizontal. The rotation angle θ(t) of the mechanical legs changes to:
[0030] θ(t)=90°-ω1t
[0031] Step 5: After the mechanical legs return to their original positions, the tracks stop rotating, allowing the robot to return to its flat ground inspection posture.
[0032] Furthermore, it also includes a robot that keeps the load-bearing surface of the stairs horizontal and descends the stairs in a reverse manner according to the reverse process of the stair climbing motion method. In the specific implementation process, the robot uses the reverse motion process to go down the stairs. It uses the rear-end depth camera to perceive the environment and stair information, and then the mechanical legs rotate forward so that the track chassis contacts the stair surface. At the same time, by adjusting the length of the telescopic rod, the robot shell is kept horizontal. The robot slowly descends on the stairs in a reverse manner to ensure smooth descent, and finally returns to the posture of flat ground inspection.
[0033] Compared with the prior art, the present invention has the following beneficial effects:
[0034] 1. This paper innovatively proposes a dual-mode method for controlling the robot to walk on flat ground and stairs. When walking on flat ground, the robot moves by relying on the wheels of the mechanical legs and the auxiliary wheels at the rear; when encountering stairs, the mechanical legs rotate to a specific angle, switch to the crawler walking mode, and rely on the telescopic rod to adjust the robot shell so that it always remains horizontal during movement. This method enables the robot to adapt to different terrains more flexibly, significantly improving the efficiency and scope of inspection tasks.
[0035] 2. This paper innovatively proposes a method to keep the main load-bearing surface of the robot horizontal during the process of climbing stairs. By precisely controlling the movement of the tracks and mechanical legs and coordinating the telescopic rod, the main body of the robot is ensured to remain horizontal throughout the entire stair climbing process. This method improves the stability and safety of the robot during stair climbing, prevents failures caused by tilting or tipping, and ensures the smooth progress of the inspection task.
[0036] 3. This paper innovatively proposes a method to identify the number and position of stairs using computer vision and deep learning methods. By acquiring images and depth information through front-end and back-end depth cameras and combining deep learning algorithms, the robot can accurately identify the edges and key points of the stairs and calculate the actual coordinates of the stairs. This method enables the robot to autonomously adjust its position and posture and accurately complete the task of going up and down stairs, significantly improving the autonomy and work efficiency of the inspection robot. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] The present invention can be more deeply understood by combining the description of the following drawings. The drawing scale of the components in the figures is not in accordance with the actual scale, but focuses on showing the principles of the embodiments. In different views, the same reference numerals are used to clearly indicate corresponding parts.
[0038] Figure 1 This is a schematic diagram of the front structure of a robot capable of maintaining a horizontal load-bearing surface when going up and down stairs according to the present invention;
[0039] Figure 2 This is a schematic diagram of the rear structure of a robot capable of maintaining a horizontal load-bearing surface when going up and down stairs according to the present invention;
[0040] Figure 3 This is a schematic diagram of the crawler chassis structure of a robot capable of maintaining a horizontal load-bearing surface when going up and down stairs according to the present invention;
[0041] Figure 4 This is a schematic diagram of the mechanical leg structure of a robot capable of maintaining a horizontal load-bearing surface when going up and down stairs according to the present invention;
[0042] Figure 5 It is a schematic diagram of the process of obtaining depth image information in a robot for maintaining the horizontal load-bearing surface of stairs going up and down and a control method thereof according to the present invention;
[0043] Figure 6 A schematic flow chart of a robot control method for maintaining the horizontal load-bearing surface of stairs going up and down according to the present invention;
[0044] Figure 7 A schematic flow chart of a robot stair climbing method for keeping the load-bearing surface of stairs level when going up and down stairs according to the present invention;
[0045] Description of Reference Numerals
[0046] 101-robot shell; 102-front-end depth camera; 103-rear-end depth camera; 3-track chassis; 4-mechanical legs; 201-rear wheel fixing bracket; 202-small auxiliary wheel; 203-hinge; 204-telescopic rod; 301-drive motor; 302-driving wheel; 303-track; 304-load-bearing wheel; 305-bottom plate; 306-support shaft; 307-guard plate; 401-leg motor; 402-leg; 403-wheel motor; 404-wheel. DETAILED DESCRIPTION
[0047] In order to more clearly illustrate the purpose, technical solutions and advantages of the present invention, the present invention is further described in detail below in conjunction with its embodiments. It should be noted that the specific embodiments described herein are only used for the explanation of the present invention and do not have a limiting effect on the present invention. For those skilled in the art, other systems, methods and / or features of the present embodiment will become apparent after reviewing the following detailed description. All of these additional systems, methods, features and advantages are deemed to be included in this specification, are within the scope of the present invention, and are protected by the appended claims. In the disclosed embodiments described in detail below, additional features will become apparent.
[0048] The present invention provides a technical solution, referring to Figure 1 and Figure 2 The robot shown in the figure can maintain the horizontal load-bearing surface when climbing stairs, comprises a robot housing (101), a crawler chassis (3) is arranged at the lower part of the robot housing (101), a rotatable mechanical leg (4) is arranged on each side of the robot housing (101), and a front end depth camera (102) and a rear end depth camera (103) are arranged at the front and rear ends of the robot housing (101) respectively for observing the environment and identifying the parameters of the stairs; a rear wheel fixing bracket (201) is fixed behind the crawler chassis (3), and the rear wheel Two small auxiliary wheels (202) are connected to both sides of the fixed bracket (201) through bearings. When the robot is running on flat ground, the robot moves in the form of a small vehicle using the mechanical legs (4) and the small auxiliary wheels (202). The crawler chassis (3) is connected to the telescopic rod (204) through a hinge (203), and the telescopic rod (204) is also connected to the robot housing (101) through the hinge (203). The length of the telescopic rod (204) can be adjusted at any time according to the posture of the robot to keep the load-bearing surface of the robot horizontal.
[0049] Reference Figure 3The crawler chassis (3) is symmetrically designed, comprising two driving motors (301), two crawlers (303) made of high-performance wear-resistant materials, and a bottom plate (305) with a thickness of 5 mm. The driving motor (301) directly drives the driving wheel (302). Each crawler (303) has three load-bearing wheels (304) supporting the crawler (303). The load-bearing wheels (304) on both sides are connected by corresponding three supporting shafts (306), and the supporting shafts (306) are fixed on the bottom plate (305). A guard plate (307) is fixed on the outer side of each load-bearing wheel (304).
[0050] Reference Figure 4 Each mechanical leg (4) includes a leg motor (401) located at the top, a rigid leg (402), a wheel motor (403) located at the bottom, and a wheel (404). The robot can use the leg motor (401) to control the rotation of the mechanical leg (4). When the robot moves on flat ground, the mechanical leg (4) remains in a vertical state, and the robot relies on the wheel motor (403) to drive the wheel (404) to move. At this time, the wheel (404) and the small auxiliary wheel (202) serve as the robot's movement mode; once facing a staircase scene, the mechanical leg (4) relies on a precise rotation drive device to adjust the angle according to the stair steps and rotate downward to an appropriate position.
[0051] Reference Figure 5 The schematic diagram of the process of capturing images by a robot with the load-bearing surface of stairs kept horizontal is shown, and the specific implementation method of the front-end depth camera (102) capturing images and predicting the position and number of stairs is as follows: First, the front-end depth camera (102) is used to obtain image and depth information. The front-end depth camera (102) will continuously collect data, record the image and depth data of the current frame, and then remove noise and outliers through filter processing, smooth the depth map, and ensure the accuracy and stability of the depth information. The filter can be applied using a Gaussian filter, and its formula is as follows:
[0052]
[0053] Among them, σ represents the standard deviation of the filter, and (x, y) represents the pixel position. Through these processing steps, high-quality images and depth information are obtained, providing a reliable data basis for the subsequent prediction of stair positions and number of steps.
[0054] Next, the image acquired by the front-end depth camera (102) is processed using the trained deep learning model (YOLO V8) to identify the location and category of the stairs in the image. The real coordinates of the stairs are calculated from the image. First, the edge points of the stairs are identified in the image. The coordinate transformation algorithm is used to transform the two-dimensional pixel coordinates into three-dimensional real coordinates using the internal and external parameters of the front-end depth camera (102). The specific transformation process is as follows:
[0055]
[0056] Where u and v are image coordinates, f x and f y are the focal length of the front-end depth camera (102), c x and c y is the coordinate of the center point of the image, X c , Y c , Z c are coordinates in the world coordinate system.
[0057] Then, computer vision technology is used to calculate the number of steps of the stairs. First, the Canny edge detection algorithm is applied to identify the edge of the stairs in the image. The specific process is as follows: apply a Gaussian filter to smooth the image and remove noise; calculate the amplitude and direction of the image gradient; apply non-maximum suppression to refine the edge; apply double threshold detection to determine the real and potential edges; and finally connect the edges through hysteresis edge tracking. After identifying the edge of the stairs, the Hough transform algorithm is used to identify the edge of each step of the stairs. The specific formula of the Hough transform is as follows:
[0058] ρ = x cos θ + y sin θ (3)
[0059] Among them, ρ is the vertical distance from the origin to the line, and θ is the angle between the line and the horizontal axis. Through the voting accumulation process, the straight lines in the image are identified, and these straight lines correspond to each step of the stairs. The number of parallel lines identified is counted, which is the number of steps of the stairs. This information is used to plan the path of the robot going up the stairs, so that it can accurately align each step of the stairs and successfully complete the task of going up and down the stairs.
[0060] Reference Figure 6 The robot control method for keeping the load-bearing surface of stairs level first obtains the RGB image and depth information of the scene through the front-end depth camera (102). The obtained image is processed using the YOLO V8 algorithm to identify the stairs and obtain the boundary box coordinates of the identified stairs.
[0061] Combining the depth information of the robot position and the stair position, a coordinate transformation algorithm is used to transform the two-dimensional pixel coordinates into the stair position in the world coordinate system with the robot as the origin. The three-dimensional position of the stair in the real space can be accurately calculated by calibrating the internal and external parameters of the front-end depth camera (102), including the number of steps and the height and width of the steps.
[0062] After determining the location and parameters of the stairs, the robot adjusts its position and moves to the front of the stairs. Then, the robot executes the stair climbing motion method.
[0063] When the robot moves to the front of the stairs, its mechanical legs (4) begin to rotate backwards, and at the same time, the wheels (404) also rotate to push the robot forward. As the mechanical legs (4) continue to rotate, the track chassis (3) gradually contacts the steps of the stairs, so that the robot enters the state of climbing stairs. In this process, the telescopic rod (204) is used to support the robot housing (101) to ensure that it remains horizontal. Once the mechanical legs (4) rotate to a position parallel to the track chassis (3), they stop rotating to prevent the legs from getting stuck on the stairs. Subsequently, the track (303) starts to rotate under the drive of the driving wheel (302), pushing the robot to gradually climb the stairs. When the robot is about to reach the last two steps of the stairs, the mechanical legs (4) return to their original positions, and the track (303) stops rotating, so that the robot returns to a flat ground inspection posture.
[0064] After completing the stair climbing task, the robot performs inspection tasks on that floor, during which it can perform tasks such as environmental monitoring, hazard warning, and object handling.
[0065] Reference Figure 7 The flowchart of a method for a robot to climb stairs while keeping the load-bearing surface of stairs horizontal is shown, and the control method includes the following steps:
[0066] First, when the robot moves to the front of the stairs, the mechanical leg (4) rotates backwards, and at the same time, the wheel (404) rotates to push the robot forward. As the mechanical leg (4) rotates, the crawler chassis (3) contacts the stair surface and changes into a stair climbing posture. The robot housing (101) is supported by the telescopic rod (204) to maintain horizontality. Specifically, the rotation angle θ(t) of the mechanical leg (4) changes with time as follows:
[0067] θ(t)=θ0+ω1t (4)
[0068] Where θ0 is the initial angle of the robotic leg (4), ω1 is the rotation speed of the robotic leg (4);
[0069] When the mechanical leg (4) rotates to the specified position, the mechanical leg (4) stops rotating to prevent the leg from getting stuck on the stairs. At this time, the rotation angle θ satisfies:
[0070] θ=90° (5)
[0071] The crawler (303) continues to rotate, pushing the robot to climb the stairs step by step. At this stage, the rotation speed of the crawler is adaptively adjusted according to the slope of the stairs and the robot load, and the optimal climbing efficiency is maintained through the PID control algorithm. The relationship between the linear speed v of the crawler and the rotation speed ω2 is:
[0072] v=r·ω2 (6)
[0073] Wherein r is the radius of the driving wheel (302);
[0074] When the robot is about to reach the last two stairs, the mechanical leg (4) starts to rotate again and gradually returns to the original position. At the same time, the length of the telescopic rod (204) is adjusted to keep the robot horizontal. The rotation angle θ changes to:
[0075] θ(t)=90°-ωt (7)
[0076] After the mechanical legs (106) are restored to their original positions, the crawler belts (303) stop rotating, so that the robot returns to a flat ground inspection posture and continues to perform the inspection task.
[0077] As the robot climbs the stairs, the system's built-in accelerometer and gyroscope monitor its stability in real time. If a tilt or sliding risk is detected, the system will automatically adjust the track speed or stop moving to prevent slipping. In addition, the robot uses a front-end depth camera (102) and other sensors to obtain real-time information about the surrounding environment and fine-tune the speed and angle of the track through sensor feedback.
[0078] In the process of climbing stairs, the robot also needs to adjust itself according to real-time environmental information. For example, when it detects a change in step height or surface irregularities, the system will adjust the rotation angle of the mechanical legs and the speed of the tracks to adapt to the new step conditions. If a failure occurs on the stairs, the robot will initiate a self-correction procedure, including reverse movement to readjust the position or send an alarm signal to the remote operator for timely human intervention.
[0079] Through these detailed control steps and supplementary measures, the robot can flexibly and safely switch from ground mode to stair climbing mode, maintain stable movement on the stairs, and finally return to ground inspection mode.
[0080] The robot's movement mode when going down stairs is opposite to the process of climbing stairs. When executing the specific movement of going down stairs, the robot adopts the reverse operation process of climbing stairs. It uses the depth camera (103) at the rear end to perceive the environment and stair information, and then the mechanical legs (4) rotate forward so that the crawler chassis (3) contacts the stair surface. At the same time, by adjusting the length of the telescopic rod (204), the robot shell (101) is kept horizontal. The robot slowly descends on the stairs in a backward manner to ensure that it can smoothly complete the action of going down stairs and finally return to the posture of flat ground inspection.
[0081] The embodiments described in this specification are only typical examples of the present invention and should not be regarded as limiting the scope of protection of the present invention. The present invention is intended to cover all modifications, equivalent substitutions and improvements that are consistent with its spirit and legal principles, which should be regarded as part of the scope of protection of the present invention.
Claims
1. A robot for keeping the load-bearing surface level when going up and down stairs, characterized in that: The robot comprises a robot housing (101), a crawler chassis (3) is arranged at the lower part of the robot housing (101), a rotatable mechanical leg (4) is arranged on each side of the robot housing (101), a front-end depth camera (102) is arranged at the front end of the robot housing (101), and a rear-end depth camera (103) is arranged at the rear end; A rear wheel fixing bracket (201) is fixed behind the crawler chassis (3), two small auxiliary wheels (202) are connected to the two sides of the rear wheel fixing bracket (201) via bearings, the crawler chassis (3) is connected to a telescopic rod (204) via a hinge (203), and the telescopic rod (204) is connected to the robot housing (101) via a hinge (203); The crawler chassis (3) is symmetrically designed, comprising two driving motors (301), two crawlers (303) and a bottom plate (305), wherein the driving motor (301) drives the driving wheel (302), each crawler (303) has three load-bearing wheels (304), the load-bearing wheels (304) on both sides are connected by corresponding three supporting shafts (306), the supporting shafts (306) are fixed on the bottom plate (305), and a guard plate (307) is fixed on the outer side of each load-bearing wheel (304); The mechanical leg (4) comprises a leg motor (401), a rigid leg (402), a wheel motor (403) and a wheel (404).
2. A robot control method for maintaining the horizontal load-bearing surface of stairs, characterized in that: Use the following steps: Step S1: Acquire images and depth information through a front-end depth camera (102); Step S2: using a deep learning algorithm to identify stairs; Step S3: Calculate the real coordinates of the stairs through the image. First, identify the edge points of the stairs in the image, and use the internal and external parameters of the front-end depth camera (102) to convert the two-dimensional pixel coordinates into three-dimensional real coordinates using a coordinate transformation algorithm. The specific conversion process is as follows: Where u and v are image coordinates, f x and f y are the focal length of the front-end depth camera (102), C x and C y is the coordinate of the center point of the image, X ω , Z ω , Y ω are coordinates in the world coordinate system; Step S4: The car moves to the front of the stairs. The positioning algorithm and the depth information obtained by the front-end depth camera (102) are used to control the car to move precisely to the front of the stairs. Step S5: Calculate the number of stairs using computer vision technology; Step S6: Execute the stair climbing exercise method; Step S7: Complete the stair climbing exercise and perform the task on this floor.
3. In the robot control method for keeping the load-bearing surface of stairs horizontal according to claim 2, the step S1 of acquiring complete image information is characterized in that: Step S1.1: using a front-end depth camera (102) installed at the front end of the robot to collect image data; Step S1.2: Record the image and depth data of the current frame; Step S1.3: Scan each pixel in the image and analyze the image to detect whether there is missing depth information (hole points) or blurred areas; Step S1.4: If all pixels have valid depth values and the image clarity reaches a preset standard, stop image acquisition; Step S1.5: If the depth information or clarity is insufficient, wait for the next frame of image, use the optical flow algorithm to track the pixels, map the depth information corresponding to the next frame to the hole points of the previous frame, and supplement the missing depth information; Step S1.6: During the image completion process, ensure that the time interval between frames is short enough to maintain the continuity and consistency of motion; Step S1.7: Repeat the above steps until all the holes are filled and the image clarity meets the standard, and then end the image acquisition process; Step S1.8: Remove noise and outliers through filter processing, smooth the depth map, and ensure the accuracy and stability of the depth information.
4. In the robot control method for maintaining the horizontal load-bearing surface of stairs as described in claim 2, the step S2 performs stair recognition, characterized in that: Step S2.1: Collect an image dataset containing ascending stairs and descending stairs, and annotate them; Step S2.2: Use the data set to train the YOLO V8 network and adjust the algorithm's built-in parameters and network architecture to improve recognition accuracy; Step S2.3: Use the trained YOLO V8 network model to process the image acquired by the front-end depth camera (102) to identify the location and category of the stairs in the image.
5. In the robot control method for keeping the load-bearing surface of stairs horizontal according to claim 2, the step S5 performs stair step detection, characterized in that: Step S5.1: Edge detection: Use algorithms such as Canny edge detection to identify the edges of the stairs; Step S5.2: Hough transform: identifying the edge of each step of the stairs by using the Hough transform algorithm; Step S5.3: Calculate the number of steps: Count the number of parallel lines identified by the Hough transform, which is the number of steps of the staircase.
6. In the robot control method for keeping the load-bearing surface of stairs horizontal according to claim 2, the step S6 performs stair climbing motion, characterized in that: The specific steps of the robot's stair climbing motion include: Step S6.1: When the robot moves to the front of the stairs, its mechanical legs (4) begin to rotate backwards, and at the same time, the wheels (404) also rotate to push the robot forward. As the mechanical legs (4) continue to rotate, the track chassis (3) gradually contacts the stair surface, allowing the robot to enter the posture of climbing stairs. During this process, the telescopic rod (204) is used to support the robot shell (101) to ensure that it remains horizontal. The rotation angle θ(t) of the mechanical leg (4) changes with time t as follows: θ(t)=θ0+ω1t; Where θ0 is the initial angle of the robotic leg (4), ω1 is the rotation speed of the robotic leg (4); Step S6.2: When the mechanical leg (4) rotates to a position parallel to the crawler chassis (3), the mechanical leg (4) stops rotating to prevent the leg from getting stuck on the stairs. At this time, the rotation angle θ satisfies: θ = 90°; Step S6.3: The crawler (303) starts to rotate under the drive of the driving wheel (302), pushing the robot to climb the stairs step by step. The relationship between the linear velocity v of the crawler (303) and the rotational velocity ω2 is: v = r·ω2; Wherein r is the radius of the driving wheel (302); Step S6.4: When the robot is about to reach the last two steps, the mechanical leg (4) starts to rotate again and gradually returns to the original position. At the same time, the length of the telescopic rod (204) is adjusted to keep the robot housing (101) horizontal. The rotation angle θ(t) of the mechanical leg (4) changes to: θ(t)=90°-ω1t; Step S6.5: After the mechanical leg (4) returns to its original position, the crawler (303) stops rotating, so that the robot returns to a flat ground inspection posture.
7. A robot control method for keeping the load-bearing surface of stairs horizontal according to claim 2, characterized in that: The movement method for going down stairs is opposite to the movement process of the robot climbing stairs in claim 6. When implemented, the robot moves according to the reverse process, uses the rear-end depth camera (103) to perceive the environment and stair information, the mechanical legs (4) rotate forward, the crawler chassis (3) contacts the stair surface, and at the same time adjusts the length of the telescopic rod (204) to keep the robot shell (101) horizontal, and descends on the stairs in a backward manner, ensuring that the robot goes down the stairs smoothly and returns to a flat ground inspection posture.
Citation Information
Patent Citations
Preparation method of solution-type acrylate adhesive
CN102477268A
Multi-legged robot based on three-degree-of-freedom mechanical legs
CN109178137A
Synergetic crawler-type four-foot robot dog
CN109927808A
Omnibearing horizontal posture stair climbing robot mechanical system and method
CN111232081A
Wheel-track combined type mobile stair climbing robot
CN112722099A
Cited By
Foot structure of humanoid robot
CN119568308A
