Mobile walking aid robot for patients with lower limb muscle weakness and control method thereof

Abstract

The application belongs to the technical field of rehabilitation medical auxiliary, discloses a mobile walking aid robot for patients with lower limb muscle weakness and a control method thereof, and is used for solving the problems of pedestrian following and obstacle avoidance control of the mobile walking aid robot in an unstructured obstacle environment, and enabling the mobile walking aid robot to keep the visibility of the followed target while ensuring a certain following accuracy and avoiding collision with obstacles. A new pedestrian trajectory prediction control method is designed to solve the problem of the detour behavior of the predicted pedestrian in the unstructured obstacle environment, and a new obstacle avoidance and visibility maintenance constraint form is designed, which can keep good theoretical properties while having a relatively simple mathematical form. Then, a cost constraint system related to pedestrian following and obstacle avoidance is established, and the control target of pedestrian following and obstacle avoidance is realized based on the rolling optimization strategy of the model predictive control framework.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of rehabilitation medical auxiliary technology, and more particularly relates to a mobile walking aid robot for patients with reduced lower limb muscle strength and a control method thereof. BACKGROUND

[0002] With the aggravation of population aging, the high incidence of motor disability in the elderly population superimposed on the lack of medical nursing labor has become a great challenge. Clinical medical data shows that at least 70% of patients can recover part of the motor function through timely and effective treatment and rehabilitation training. In order to encourage patients with mild and moderate lower limb muscle weakness to actively participate in rehabilitation training and alleviate the shortage of rehabilitation physicians, it is necessary to develop a mobile walking aid robot with power support and follow-up monitoring functions that can be used flexibly in indoor and outdoor scenes. As an aid for patients with reduced lower limb muscle strength, the mobile walking aid robot in the unstructured scene will be disturbed by static or even dynamic obstacles when it is in the follow-up monitoring state. It is a great challenge to keep collision-free and monitor the motion state of the person at all times in this scenario, so that the walking aid robot can provide timely assistance to the person.

[0003] Specifically, the above problems can be summarized as the trajectory prediction problem of the walking aid robot for the person (patient with reduced lower limb muscle strength) and the visibility problem of the follow-up target. For the trajectory prediction problem of the walking aid robot for the person, it is found through research that the person generally prefers a "non-direct" obstacle avoidance method, that is, the person starts to avoid obstacles at a relatively close distance, and in a relatively narrow obstacle environment, the person also has to adopt this obstacle avoidance strategy, that is, the person starts to avoid obstacles at a relatively close distance. This means that after the person completes the obstacle avoidance action, there is generally only a small time window left for the robot to respond, and for turning prediction, there is also a lack of strong correlation features before approaching the obstacle, which brings great uncertainty to the robot for calculating the turning direction, and there is a problem of predicting the turning behavior of the person in the unstructured obstacle environment.

[0004] The visibility of the following target is a critical issue. Due to the limited detection angle of sensors and complex obstacle environments, the pedestrian may leave the field of view of the assistive robot or be obstructed by obstacles, leading to a decrease in the quality of the robot's perception data and even target loss. Existing algorithms generally improve the visibility of the following pedestrian within the robot's sensors by increasing the distance from the obstacle to the line of sight between the robot and the obstacle, but this lacks corresponding theoretical guarantees, resulting in low reliability. Therefore, modeling this problem and giving it sound theoretical properties to improve the reliability of the assistive robot is a significant challenge. Furthermore, pedestrian following control needs to simultaneously consider following accuracy, obstacle avoidance planning, and the visibility of the following target, and a trade-off among these three factors is necessary to ensure successful following. However, comprehensive research on this motion planning problem is still lacking. Summary of the Invention

[0005] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention provides a mobile walking assistance robot and its control method for patients with lower limb muscle weakness. The purpose is to solve the problem of predicting the detour behavior of pedestrians in unstructured obstacle environments and improve the reliability of pedestrian following and obstacle avoidance control of mobile walking assistance robots in unstructured obstacle environments.

[0006] To achieve the above objectives, the present invention provides a mobile walking robot control method for patients with lower limb muscle weakness, comprising: predicting the walking path of the target followed by the mobile walking robot in the next N time steps, and controlling the mobile walking robot based on the walking path to perform target following control;

[0007] Predicting the walking path specifically includes:

[0008] Step 1: Place the target in The predicted position at any given moment is used as a node in a binary tree, and this node is used as the current node to determine the desired direction of movement in an accessible environment. Is this a feasible direction to travel? If so, then proceed. directional extension distance If the target is not the root node, proceed to step three if the current node is the root node, and proceed to step two if the current node is not the root node. The target's position at the initial moment is the root node of the binary tree.

[0009] Step 2: Determine if the direction extended by the current node's parent node is a feasible walking direction. If so, extend the path along the direction extended by the parent node. Reach the child node of the current node, where the direction in which the parent node extends is the direction in which the target is located. Predict the direction of motion at any given moment; if not, proceed to step three.

[0010] Step three: taking the current desired moving direction as the center, judging whether there is a turnable direction on the left and right sides of the current node according to the turn angle from small to large, wherein the turn angle does not exceed a preset first turn angle Step four: taking the current desired moving direction as the center, judging whether there is a turnable direction on the left and right sides of the current node according to the turn angle from small to large, wherein the turn angle is between a preset first turn angle and a preset second turn angle Step five: setting the child node of the current node to represent a stop moving state, and the extension direction of the child node is the same as the current node, and the extension distance is 0. Step six: taking the child node of the current node as the predicted position at the time t, and letting t = t + 1, entering the iterative search process of steps one to five until N times of iterative search are reached, obtaining a binary tree representing the predicted target walking path, and each branch of the binary tree representing a possible walking path. Further, it further includes constructing a convex polygon safety area at the time t, and the convex polygon safety areas corresponding to each time on each branch of the binary tree are connected in series to form a safe corridor without obstacles and obstructions as a passable area for the mobile walking aid robot in the following control process of the target; wherein the convex polygon safety area at the time t is constructed, including:

[0011]

[0012]

[0013] Step six: taking the child node of the current node as the predicted position at the time t, and letting t = t + 1, entering the iterative search process of steps one to five until N times of iterative search are reached, obtaining a binary tree representing the predicted target walking path, and each branch of the binary tree representing a possible walking path.

[0014] Further, it further includes constructing a convex polygon safety area at the time t, and the convex polygon safety areas corresponding to each time on each branch of the binary tree are connected in series to form a safe corridor without obstacles and obstructions as a passable area for the mobile walking aid robot in the following control process of the target; wherein the convex polygon safety area at the time t is constructed, including:

[0015] taking the predicted target position at the time t as the center ​​​​​​​​​​​​​​, the center is divided into sectors with equal angular intervals , , is a preset experience value

[0016] , the projection of each obstacle in the th sector to the angle bisector of the th sector and take the minimum value :

[0017]

[0018] , the unit vector of the th sector , representing the direction of the angle bisector of the th sector , represents the transpose , the angle between the angle bisector of the th sector and the x-axis in the world coordinate system , the set of visible obstacles in the world coordinate system , the set of visible obstacles in the th sector , any visible obstacle in the th sector , the maximum distance from the center of the given safety area to the th sector

[0019] The convex polygon safety area obtained at the th moment is represented by the following set:

[0020]

[0021] , wherein , , .

[0022] Further, it also includes expanding the convex polygon safety area, including:

[0023] Translate the edges of the convex polygon safety area outward in order, and the translation principle is: if the th edge and other edges have common excluded obstacles, then move the th edge to the obstacle corresponding to the projection minimum value excluded only by the th edge

[0024] If the ​After the obstacle is eliminated by the first edge and other edges, there is no obstacle that is eliminated by the first edge alone. If an obstacle is eliminated by an edge, then the first edge will be eliminated. The strip is shifted outwards to At this time, the place Strip edge distance from center The maximum distance is .

[0025] Furthermore, in step one, the desired direction of movement in the barrier-free environment is determined. Whether it is a feasible direction of travel includes:

[0026] If the current desired direction of movement Falling sector satisfy ,but The feasible directions of travel; among them, For sectors Internal obstacles and Connect the predicted target location to the sector The minimum value of the projection of the angle bisector. Preset empirical coefficients;

[0027] In step two, it is determined whether the direction extended by the parent node of the current node is a feasible walking direction, including:

[0028] The goal is The sector into which the motion direction falls at any given moment satisfy If the current node's parent node extends in a certain direction, then that direction is a feasible walking direction; where, For sectors Internal obstacles and Connect the predicted target location to the sector The minimum value of the projection of the angle bisector;

[0029] In step three, the desired direction of movement is... Based on the center, judge step by step according to the detour angle from small to large. Are there any alternative routes to the left and right, including:

[0030] By sector Scanning from the center outwards gradually to both sides. Grid sector, , Represents the preset first detour angle; if the first The sector the step falls into satisfy Then in If there is a circumventable direction on the left, end the left-side scan; if the first... the sector where the step falls satisfies , there is a detourable direction on the right side of the step, ending the right side scanning; wherein, , , , respectively represent the minimum value of the line connecting the obstacle in the sector , and the predicted target position at the moment to the projection of the angle bisector of the corresponding sector; the detour angle of the detourable direction is the angle bisector direction of the sector with the minimum detour angle;

[0031] In step four, the current expected moving direction is taken as the center to gradually determine whether there is a detourable direction on the left and right sides of the step according to the detour angle from small to large, including: The sector

[0032] is taken as the center, and scanning is performed from the first th sector to the first th sector on both sides, , , characterize the preset second detour angle; if the sector where the step falls satisfies , there is a detourable direction on the left side of the step, ending the left side scanning; if the sector where the step falls satisfies , there is a detourable direction on the right side of the step, ending the right side scanning; wherein, . Further, the approximation degree of each possible walking path in the binary tree to the planned path of the mobile walking aid robot at the current moment is calculated , the walking path corresponding to the minimum is selected as the target most possible walking path, and the mobile walking aid robot is controlled based on the target most possible walking path to follow the target for control; wherein,

[0033]

[0034]

[0035] In the formula, represents the current moment, represents the planned path of the mobile walking aid robot at the last moment.

[0036] ​​​​​Further, the mobile walking aid robot is controlled based on the most possible walking path of the target to perform following control on the target, including:

[0037] The field of view cost and the field of view constraint of the mobile walking aid robot are determined based on the predicted position of the target at each time in the walking path, the safety corridor constraint of the mobile walking aid robot is constructed based on the safety corridor corresponding to the walking path, the motion amplitude constraint, the field of view constraint and the safety corridor constraint of the mobile walking aid robot are converted into corresponding quadratic penalty functions;

[0038] A finite time domain optimal control problem is constructed with the sum of the following cost as an objective function, including the motion model constraint of the mobile walking aid robot, and the control variables of the mobile walking aid robot are solved to control the following motion of the mobile walking aid robot, including the following cost, the obstacle avoidance cost, the field of view cost and the quadratic penalty functions;

[0039] The following cost is:

[0040]

[0041] In the formula, represents the actual position error and the expected position error of the mobile walking aid robot, and the actual linear velocity error and the expected linear velocity error; represents the actual linear acceleration error and the expected linear acceleration error of the mobile walking aid robot; , are the forward and lateral position error components respectively, , are the forward and lateral linear velocity errors respectively, , are the forward and lateral linear acceleration error components respectively; , , , , , are the weights of the corresponding items;

[0042] The safety corridor constraint is:

[0043]

[0044] In the formula, is the actual position of the mobile walking aid robot at time, is the minimum safety distance between the mobile walking aid robot and the obstacle.

[0045] Further, the quadratic penalty function corresponding to the motion amplitude constraint is:

[0046]

[0047] In the formula, To constrain the amplitude of the linear velocity and angular velocity of the mobile walking robot, To constrain the amplitude of linear and angular acceleration of the mobile walking robot; As a preset public weight, Private weights; , , , The maximum linear velocity, linear acceleration, angular velocity, and angular acceleration of the mobile mobility assistance robot; , , , Mobile walking assistance robots The actual velocity, acceleration, angular velocity, and angular acceleration at any given moment; , , , These are the weighting coefficients for linear velocity, angular velocity, linear acceleration, and angular acceleration, respectively. This is for taking the absolute value; This is an operation to obtain the 2-norm;

[0048] The quadratic penalty function corresponding to the field of view constraint for:

[0049]

[0050] In the formula, for The walking robot follows the target at the angle relative to the center line of the field of view. Human-machine distance, To half the field of view of the vision sensor of the mobility aid robot, , The upper and lower limits of human-machine distance; , These are the weighting coefficients corresponding to the field of view and the human-machine distance, respectively.

[0051] Quadratic penalty function corresponding to safe corridor constraint for:

[0052]

[0053] In the formula, These are the weighting coefficients of the quadratic penalty function corresponding to the safety corridor constraint.

[0054] The present invention also provides a mobile assistive robot controller for patients with lower limb muscle weakness, including a computer-readable storage medium and a processor;

[0055] The computer-readable storage medium is used to store executable instructions;

[0056] The processor is used to read executable instructions stored in the computer-readable storage medium and execute the mobile assistive robot control method for patients with lower limb muscle weakness described above.

[0057] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the mobile assistive robot control method for patients with lower limb muscle weakness as described in any of the preceding claims.

[0058] The present invention also provides a mobile assistive robot for patients with lower limb muscle weakness, characterized in that it includes a controller, the controller being used to execute the mobile assistive robot control method for patients with lower limb muscle weakness described above, so as to perform target following control; or the controller is the mobile assistive robot controller for patients with lower limb muscle weakness described above.

[0059] In summary, the above-described technical solutions conceived in this invention can achieve the following beneficial effects:

[0060] (1) In order to solve the problem of predicting pedestrian detour behavior in unstructured obstacle environment, this invention models the pedestrian detour behavior in unstructured obstacle environment. Specifically, a five-stage path search method is designed to simulate the pedestrian detour decision and realize the prediction of pedestrian detour trajectory by mobile walking robot in this environment. At the same time, considering the uncertainty of detour behavior, a binary tree structure is used to give multiple possible detour directions of pedestrians, which improves the reliability of control method in complex unstructured environment.

[0061] (2) Furthermore, the present invention designs a safe corridor constraint with a simple mathematical form and excellent theoretical properties, constructs a convex polygon safe region at each time, and connects the convex polygon safe regions at each time to form an unobstructed and unblocked safe corridor as the passable area of ​​the mobile walking robot. It proposes to use the basic properties of convex sets to theoretically ensure the visibility of the walking robot following the pedestrian during the pedestrian following process and avoid collisions with environmental obstacles.

[0062] (3) As a preferred option, based on the principle that if an obstacle is excluded by one edge of the convex polygon safety area, it is already outside the convex polygon area and does not need to be excluded by the other edges, the convex polygon safety area is expanded to avoid excluding too much unobstructed space.

[0063] (4) Furthermore, this invention designs a pedestrian following and obstacle avoidance control method based on a safe corridor, transforming the control task into a series of finite-time domain optimization problems containing model constraints of the controlled object. This allows for the convenient explicit addition of the required control costs and constraints. The constructed safe corridor constraints ensure the visibility of the following target relative to the mobile assistive robot and prevent collisions with obstacles. The constructed following cost, through the design of forward and lateral error components and their corresponding weights, enables the robot to tend to move sideways to avoid obstacles. This multi-objective optimization framework based on model predictive control can achieve pedestrian following, obstacle avoidance, and visibility maintenance of the following pedestrian target in unstructured obstacle environments, realizing smooth pedestrian following in complex unstructured environments.

[0064] (5) Furthermore, considering the inaccuracy of long-term predictions, a common weight is assigned to each penalty function. And a private weight This allows the control method of the present invention to operate according to the attenuation rate. The weight of each constraint is gradually reduced as the prediction time increases from near to far, while focusing on constraints at more recent times to prevent over-obstacle avoidance.

[0065] In summary, this invention addresses the problem of pedestrian following and obstacle avoidance control in omnidirectional mobile walking robots operating in unstructured obstacle environments. It enables the robot to maintain a certain level of following accuracy while avoiding collisions with obstacles, ensuring the visibility of the following pedestrian. This invention designs a novel pedestrian trajectory prediction control method to solve the challenge of predicting pedestrian detour behavior in unstructured obstacle environments. It also designs a new constraint form for obstacle avoidance and visibility maintenance, which maintains good theoretical properties while having a relatively simple mathematical form. Then, by establishing a cost constraint system related to pedestrian following and obstacle avoidance, and solving it based on a rolling optimization strategy within a model predictive control framework, the control objective of pedestrian following and obstacle avoidance is achieved. Attached Figure Description

[0066] Figure 1 This is a schematic diagram of the control method in an embodiment of the present invention.

[0067] Figure 2 This is a schematic diagram illustrating the construction of a secure area in an embodiment of the present invention.

[0068] Figure 3 This is a schematic diagram of the physical quantities used in this invention.

[0069] Figure 4 This is a schematic diagram illustrating the calculation of forward and lateral error components in an embodiment of the present invention. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of this invention clearer, the invention 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 merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0071] In this invention, the terms "first," "second," etc., used in the invention and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0072] Example 1

[0073] The control method in this embodiment of the invention first needs to predict the future walking path of a pedestrian. For pedestrian trajectories under obstacle-free conditions, the turning speed of the pedestrian can be predicted based on the basic assumption of angular velocity decay, that is, it is assumed that the speed of change of the pedestrian's movement direction will gradually decrease at a certain rate until the pedestrian walks in a straight line. On this basis, under obstacle-containing conditions, the influence of obstacles on the deflection of the pedestrian's walking path can be further considered. For this, it can be assumed that the pedestrian prefers to change its movement direction at the smallest possible angle. Specifically, it can be assumed that the pedestrian's detour decision will go through multiple stages. At the current moment, the pedestrian will first determine whether it can move in the desired direction in an obstacle-free environment. If so, it will move in the desired direction; if not, that is, if the direction is blocked by an obstacle, it will determine whether the movement direction at the previous moment was a feasible walking direction. If so, it will walk in the movement direction at the previous moment to ensure the continuity of the trajectory. If both of the above possibilities are rejected, the pedestrian will gradually determine the unobstructed detour directions to the left and right of the current desired direction of movement, starting from the center and moving outwards. However, the maximum detour angle in this process will be limited (within a preset first detour angle). Assuming that within the preset first detour angle, at least one of the left and right sides has a smaller feasible detour direction, the pedestrian will choose the feasible detour direction with the smallest detour angle (if the minimum feasible detour angles on both sides are the same, then both the left and right sides will be selected). If the detour angles on both sides exceed the first detour angle, but within the preset second detour angle, if there are feasible detour directions on both sides, the direction with the smallest detour angle on each side will be used as the current detour direction. If only one side has a feasible detour direction, the pedestrian will detour along the feasible detour direction with the smallest detour angle on that side. If no feasible walking direction can be found in the above process, i.e., the detour angles on both sides exceed the maximum detour angle, the pedestrian will enter a stopped state. Taking the next moment as the current moment, the search is performed for N prediction time steps according to the above process to obtain multiple possible walking trajectories. Each walking trajectory contains the predicted pedestrian position and direction of movement at N time steps. These multiple possible walking trajectories form a pedestrian trajectory binary tree, where each branch represents a possible walking trajectory.

[0074] The invention then analyzes the surrounding environment time-by-time along the predicted pedestrian trajectory and establishes a series of safe zones with convex polygonal walking patterns. Specifically, the invention uses the predicted pedestrian position at the current moment as the center, divides the two-dimensional plane (i.e., the ground) where the pedestrian position is located into multiple equal sectors, scans obstacles falling within each sector, and sets a dividing line within the corresponding sector to exclude all obstacles within that sector. That is, it calculates the... Each obstacle within a sector is connected to the center via a line. The projection of the angle bisectors of each sector is taken as the minimum value. On the angle bisector of each sector, at a distance from the center Draw a perpendicular line at each point; this perpendicular line is the dividing line within the corresponding sector. By combining the dividing lines of all sectors at the current moment, a convex polygonal safe region can be formed, which contains no obstacles. Then, by connecting the convex polygonal safe regions at each moment along the corresponding path, an unobstructed and unblocked safe corridor can be formed (each path in the pedestrian trajectory binary tree corresponds to a safe corridor), which serves as the robot's traversable area.

[0075] Based on the pedestrian trajectory prediction and safety corridor given above, a corresponding finite-time domain optimization control problem can be constructed to select the path most likely to be chosen by the pedestrian from multiple feasible paths for following. This invention constructs corresponding costs and constraints, including following cost, obstacle avoidance cost, field of view cost, as well as motion amplitude constraints, field of view constraints, and safety corridor constraints. Furthermore, this invention assumes that the omnidirectional robot (assistive robot) follows a first- or second-order motion model. Subsequently, first- and second-order penalty functions are used to remove the remaining constraints (motion amplitude constraints, field of view constraints, and safety corridor constraints) outside the robot's motion model, transforming the motion amplitude constraints, field of view constraints, and safety corridor constraints into corresponding penalty functions. Based on this, the finite-time domain optimization control problem can be solved using the BFGS quasi-Newton optimization algorithm. According to the model predictive control algorithm framework, the control variables in the first moment of the optimized control variable sequence are adopted for controlling the robot's following motion.

[0076] In this embodiment of the invention, for the construction of convex polygon safety areas: for robot path planning, safety corridors can be used to ensure that the robot runs in an obstacle-free area. However, in this embodiment of the invention, by constructing a series of specially configured convex polygon safety areas, the collision safety of the mobile assistive robot can be ensured at the same time, and the target can be avoided from being obscured by obstacles.

[0077] Specifically, a single convex polygon safe region is a convex polygon region that does not contain any obstacles. As for the calculation of this region, such as... Figure 2 As shown, a reference center point needs to be set first. In trajectory prediction, the reference center point is the predicted pedestrian position at the current moment, and the maximum distance from the safety zone boundary to the center is given. , The maximum allowable distance between the pedestrian and the mobile walking robot can be selected. Then, a reference center point is used. Centered on the plane, the plane containing the center point is divided into the following sections at equal angular intervals: a sector interval , As a preset experience value, the first Each sector interval can be represented as:

[0078]

[0079] To simplify calculations, obstacle avoidance is then considered within the robot's configuration space. Assume the robot can pass through an obstacle with a radius of... In the circular description, the robot can be considered a point mass within the configuration space, while the obstacles expand outwards from their original position. The obstacle is considered as a sector with a radius of Circle. Based on this, scan all points (obstacles) falling within each sector, and calculate the distance between each obstacle and the center in each sector. The projection of the line drawn to the angle bisector of the sector is taken as the minimum value, where the first... Each obstacle within a sector and the center Connect to the The projection of the angle bisectors of each sector is taken as the minimum value. :

[0080] (1)

[0081] in, Indicates the first The sector, the Unit vector of each sector , characterizing the The direction of the angle bisector of each sector This indicates transpose, while For the first The angle between the angle bisector of each sector and the x-axis in the world coordinate system, where the x-axis in the world coordinate system is set according to actual needs. This represents the set of visible obstacles in the world coordinate system. Indicates the first The set of visible obstacles in each sector Indicates the first Any visible obstacle in each sector. From the minimum value calculation of the outermost layer of equation (1), it can be seen that if the minimum value... If the nearest obstacle is beyond the maximum distance, then... Set as .

[0082] In the On the angle bisector of each sector, the distance from the center of Draw a perpendicular line from point 1 to point 2; this perpendicular line is the first perpendicular line. Each sector is divided by a dividing line that can eliminate all obstacles within that sector. The dividing lines of each sector are combined to form a convex polygon safe zone, in which there are no obstacles.

[0083] In this embodiment of the invention, the convex polygonal region constructed by the above method can be represented by the following set:

[0084] (2)

[0085] in:

[0086]

[0087] This method can guarantee the reference center point It must be contained within the convex polygon. Clearly, when ,and Then there must be At this point, the reference center point will be used. Substituting into the left side of inequality (2), we get:

[0088]

[0089] Therefore, the above conclusion holds. This property can be used to ensure the visibility of the following target relative to the mobile assistive robot. On the other hand, the convex polygon necessarily does not contain obstacles. This can be proven by equation (1):

[0090]

[0091]

[0092] Based on this conclusion, for The first matrix inequality (2) The action will inevitably not satisfy the condition. And it is obvious that... That is, any obstacle point It must belong to a certain interval, therefore... The matrix inequality (2) must have one row that is not satisfied, which means that the convex polygon must not contain any obstacles.

[0093] Therefore, based on the predicted pedestrian and obstacle positions at different times, a convex polygonal safe zone can be constructed for the corresponding time.

[0094] Convex polygon safety area expansion: As a further design of the present invention, considering that the safety area obtained through the above process may be small in narrow passages, resulting in the loss of a considerable amount of unobstructed space, it can be expanded a second time. In the embodiments of the present invention, in accordance with the principle of "not needing to eliminate obstacles multiple times", the principle is that as long as an obstacle does not satisfy one of the inequalities in equation (2), it is already outside the convex polygon area and does not need to be eliminated by the other inequalities. As for how to achieve this principle, the embodiments of the present invention provide a simple method to expand the convex polygon area obtained above. The edges are translated sequentially from the inside out and then outwards, following the principle that if the first edge... If one edge and the other edges share an obstacle that they can collectively eliminate, then move the first edge. The edge is only from the first At the obstacle with the smallest projected distance excluded by the edge, if the first obstacle is excluded... After the obstacle is eliminated by the first edge and other edges, there is no obstacle that is eliminated by the first edge alone. If an obstacle is eliminated by an edge, then the first edge will be eliminated. Move the edge to the maximum allowed distance For example, obstacle Q is the first position in a convex polygon region. One edge and another edge (let's assume it's the first one) If obstacles are eliminated simultaneously on both sides (+1), then the obstacle is eliminated through the first side. +1 edges are enough to remove obstacle Q, therefore, the first edge can be removed. The edge moves to only the first The obstacle with the smallest projected distance excluded by the edge is located at the point where the projected distance is the th. After removing obstacle Q within each sector, the remaining obstacles and the center Connect to the The projection of the angle bisector of each sector.

[0095] The expanded safe zone clearly still does not contain any obstacles, and since the original safe zone is a subset of the expanded safe zone, the central reference point can be guaranteed. Still within the safe zone of the convex polygon middle.

[0096] Pedestrian obstacle avoidance behavior prediction and safe corridor construction: By determining the center reference point of the convex polygon safe area at each moment at the position of the following target, the purpose of obstacle avoidance and maintaining target visibility can be achieved simultaneously. The role of maintaining target visibility can be proved by the basic properties of convex sets and the above-proven inference that the reference center point must be contained within the convex polygon.

[0097] Obviously, according to equation (2), the safe region of the convex polygon at a given time can be determined. It is a convex polygon, a type of convex set. According to the properties of convex sets, any point on the line connecting any two points within the convex set is contained within the convex set. Therefore, a known pedestrian target must be contained within the convex set. Inside, and when the robot meets the constraints, it is also in the safety corridor. Therefore, the robot's line of sight (LOS), i.e., the line of connection between the target (the patient with lower limb muscle weakness) and the robot, is completely enveloped within this area. Inside. And the above proof Since the area contains no obstacles, the LOS obviously will not intersect with any obstacles, meaning the target will not be occluded by obstacles. Furthermore, this method can guarantee visibility even if the pedestrian's position shifts due to inaccurate prediction, as long as the actual pedestrian position does not leave the set convex polygon area. Since the pedestrian's position is generally a considerable distance from the boundary of the convex polygon, this method often has significant redundancy in addressing the problem of predicted position shift.

[0098] To construct a safe corridor, it is first necessary to predict the walking trajectory of the target being followed. In an obstacle-free environment, this embodiment of the invention proposes prediction based on the assumption of angular velocity decay. If a pedestrian... Location at any time Speed ​​magnitude Orientation (direction of movement) and the angular velocity of the direction. Then, there is a pedestrian motion prediction model in an obstacle-free environment:

[0099] (3)

[0100] in, It is the attenuation factor of the pedestrian's turning angular velocity. This is the system sampling time interval. In this embodiment of the invention, The assumption based on the decay of turning angular velocity better reflects the long-term turning behavior of pedestrians, because pedestrians are unlikely to significantly change their walking direction over a prolonged period. The definitions of some pedestrian motion states are as follows: Figure 3 As shown, Figure 3 In Use the world coordinate system.

[0101] However, in obstacle-prone environments, pedestrians may be obstructed by obstacles and need to detour. To address this issue, in this embodiment of the invention, obstacle information provided by the convex polygonal safe region constructed at each moment (i.e., The control method predicts pedestrian obstacle avoidance behavior. It first sets a preset coefficient. (Empirical value) Based on the motion model of equation (3), pedestrian movement is predicted step by step, and then a safety corridor is constructed along the predicted trajectory. When there are pedestrians in the area, If a pedestrian deviates too closely from an obstacle, they will consider it too close and actively avoid it with a small deflection. Based on this, this invention proposes the following five-stage path search method to simulate a pedestrian's detour decision. First, a root node of a safety corridor binary tree is set with the pedestrian's initial position as the reference center point. Then, the search iteratively follows these steps:

[0102] Step 1: In At any given moment, determine whether movement can proceed in the desired direction within an accessible environment; if so, proceed in the desired direction. This includes calculating the current desired displacement. , target The predicted position at time t is treated as a node in the binary tree, denoted as the current node. If the movement is in the original predicted direction... The sector range to be entered (let's call it the sector range) )satisfy This means being able to move in the desired direction within an accessible environment, where... For each obstacle within sector interval a, and Time and location The minimum value of the projection of the line drawn to the angle bisector of that sector is the minimum value along the binary tree of the safety corridor. directional extension distance Reach the child node of the current node (taking the current position as the child node of the current node), and enter that child node for iterative search (taking that child node as the current node, and taking its position as the child node). The predicted location at time will (Using the current moment as the starting point, the iterative search continues according to each step), where, Magnitude of velocity at any moment and direction of movement Determine according to formula (3); otherwise ( If the current node is the root node of the binary tree, proceed to step three; otherwise, proceed to step two. Initially, the current node (the pedestrian's initial position) is the root node.

[0103] Step Two: In At any given moment, the direction of pedestrian movement Obstructed by obstacles ( When the current node's parent node (the node preceding the current node) extends in a sector (the movement direction in the previous moment), it will determine whether the previous movement direction is a feasible direction. Specifically, this includes: if the current node's parent node (the node preceding the current node) extends in a sector (defined as an interval), it will determine whether the previous movement direction falls within that sector. )satisfy This indicates that the direction of motion in the previous moment was feasible, among which, sector interval Internal obstacles and Predicted pedestrian locations in real time The minimum value of the projection of the line connecting to the angle bisector of that sector is the distance the binary tree of the safety corridor extends along the direction of movement in the previous moment. If the current node reaches its child node, proceed to that child node for iterative search; otherwise (if both possibilities in step one and step two are rejected), proceed to step three.

[0104] Step 3: Pedestrians will use their current desired direction of movement as the center and gradually determine unobstructed detour directions to the left and right of that desired direction of movement from the inside out. This specifically includes: using sector intervals... Scanning from the center outwards gradually to both sides. Grid sector, ,in, Represents the preset first detour angle, if the first... The sector that the step falls into ( or At least one sector in ) satisfies ( and / or This indicates that within the preset first detour angle, there is at least one smaller detour direction on the left and right sides. , (used to characterize the smaller traversable direction), where, Characterized by sector intervals The angle at which it moves to both the left and right simultaneously around the center. The smaller the value, the smaller the detour angle. , Represents sector interval or Internal obstacles and Time and location The minimum value of the projection of the line drawn to the angle bisector of the sector is the minimum value of the sector interval ( or (Inner) Include candidates and extend the direction and end the scan. If both sides simultaneously satisfy ( and ), then the interval and All were included in the candidates. and Represents the current minimum bypass direction; otherwise, continue scanning until... The scan ends when the condition is met. If a direction that meets the condition is found after the scan, then in all candidate sector intervals ( , or / and If the sector with the smallest detour angle is selected as the selected direction (if the smallest detour angles on both sides are the same, then both sides are selected), then the safe corridor binary tree extends a certain distance towards the angle bisector of the sector with the smallest detour angle. If the current node reaches its child node, proceed to the iterative search of all child nodes; otherwise, proceed to step four.

[0105] Step 4: Using sector intervals Centered on the first, radiating outwards to both sides from the second The grid sector scan reached the first Grid sector interval, , Characterizes the preset second detour angle, This step first scans the left side; if the first... The sector the step falls into Satisfying in direction This indicates that there is a feasible detour direction on the left, therefore the sector Include candidates and end the left-side scan. Then scan the right side, if the... The sector the step falls into Satisfying in direction This indicates that there is a feasible detour direction on the right side, therefore the sector The candidate intervals are included, and the right-side scan ends. If at least one interval satisfying the condition is found after the scan, the safe corridor binary tree extends towards the angle bisectors of all candidate intervals and enters an iterative search of all child nodes; otherwise, proceed to step five. It should be noted that in other embodiments, the right side can be scanned first, followed by the left side.

[0106] Step 5: Upon entering this stage, the pedestrian perceives their movement as being obstructed. Therefore, a unique child node is set in the safety corridor binary tree. This child node represents the stopped movement state, and its extension direction is the same as the current node, but its extension distance is 0 (i.e., the reference center point is the same as the current node). Then, the iterative search of this child node begins.

[0107] Step 6: Reach the maximum search step count N to obtain a pedestrian trajectory binary tree, where each branch represents a possible walking trajectory.

[0108] The total search depth of this method is ,in This is the number of prediction steps in the model predictive control. At this point, the depth of each path in the binary tree obtained using this method is... Each of these represents a possible pedestrian detour trajectory. However, the model predictive control algorithm only accepts one possible prediction, so it is necessary to extract the most probable path from the tree-like safety corridor to calculate the control cost.

[0109] To address this problem, this invention makes the following assumption: when there are two possible detour directions for the target being followed, and these directions are equally probable under the influence of other factors, the target will tend to choose the direction that the robot follows. Based on this, in this embodiment of the invention, the approximation between the pedestrian detour path predicted by the mobile assistive robot and the robot's planned path is calculated according to the following formula:

[0110]

[0111] in, Indicates the current moment. This indicates the path planned by the mobile mobility robot at the previous moment.

[0112] Since a new round of finite-time optimization has not yet been performed, we can only use the planning results from the robot's previous round. This serves as a reference trajectory for calculating the approximation level. Then, the trajectory with the lowest cost is selected. The smallest branch is taken as the most likely branch in the binary tree of the safety corridor, and the pedestrian position and corresponding safe area are obtained in the future N time steps predicted by the walking robot. This problem can be solved by backtracking.

[0113] Based on the pedestrian positions predicted by the walking robot at each time step, the field of view cost and field of view constraints of the walking robot are determined, based on the safe area at each time step. , Construct safe corridor constraints for mobility aid robots.

[0114] With the optimization objective of minimizing the following cost, obstacle avoidance cost, and field of view cost of the walking assistance robot, the control variable sequence of the walking assistance robot is obtained under constraints including the robot's motion amplitude constraint, field of view constraint, safe corridor constraint, initial state constraint, and motion model constraint, so as to control the mobile walking assistance robot.

[0115] In the solution process, the motion amplitude constraints, field of view constraints, and safe corridor constraints of the walking robot are converted into corresponding penalty functions. The optimization objective is to minimize the sum of the following cost, obstacle avoidance cost, field of view cost, and corresponding penalty functions of the walking robot. Under the initial state constraints and motion model constraints of the walking robot, the sequence of control variables of the walking robot is obtained to control the mobile walking robot.

[0116] One advantage of this invention is that the constrained space formed by the safe corridor is convex, which is beneficial for finding the optimal solution. Another advantage is that it does not require complex preprocessing of the visual sensor data of the walking robot; only the raw point cloud data is needed. Simultaneously, it implicitly includes an automatic distance adjustment strategy. Because the size of the safe zone shrinks when a pedestrian is in a narrow area, it forces the robot to move closer to the pedestrian. This strategy has many benefits for pedestrian following, such as reducing the possibility of losing the user in narrow areas.

[0117] In this embodiment of the invention, multiple costs and constraints are proposed, including following cost, obstacle avoidance cost, field of view cost, as well as motion amplitude constraint, field of view constraint and safe corridor constraint, to design the path planning of the mobile assistive robot at each moment.

[0118] 1. The cost of following a walking robot:

[0119]

[0120] in, This represents the actual position error and expected position error of the mobile walking robot, as well as the actual speed error and expected speed error. This represents the difference between the actual acceleration error and the expected acceleration error of the mobile walking robot. , These are the forward and lateral position error components of the mobile walking assistance robot, respectively. , These are the forward and lateral velocity errors, respectively. , These are the forward and lateral acceleration error components, respectively. Specifically, the forward position error... It is the total error Along the desired speed Projection of direction , These represent the actual position and the desired position of the mobile walking robot, respectively, while the lateral position error... That is the remaining part of the total error, which is equal to... Perpendicular to each other. A schematic diagram of the calculation process can be seen. Figure 4 The calculation method for the remaining velocity and acceleration error components is the same, which involves projecting along the desired velocity direction and the perpendicular direction. , , , , , The weights for the corresponding items are determined based on experience.

[0121] In this embodiment of the invention, when calculating the following cost, by designing forward and lateral error components and their corresponding weights, the robot is made to tend to move to the side to avoid obstacles.

[0122] 2. The cost of obstacle avoidance for mobility aid robots J obs :

[0123]

[0124] in, This represents the maximum visible distance to the obstacle, i.e., the distance from the robot to the obstacle. They are respectively The number of obstacles inside, This represents the 2-norm operation. This indicates an indicator function that, when a condition is met, immediately... If , it is 1; otherwise, it is 0.

[0125] 3. The field of view cost of a walking robot:

[0126]

[0127] in, Let be the arc length of the pedestrian relative to the center line of the field of view, based on the prediction of the walking assistance robot. The pedestrian's position at any given time is determined. The weights are determined by empirical values.

[0128] 4. Motion range constraints of the walking robot:

[0129]

[0130] in, , These represent the velocity components of the walking robot on the x and y axes in the world coordinate system. , These represent the acceleration components of the walking robot along the x and y axes in the world coordinate system. Let ω represent the angular velocity and angular acceleration of the walking robot in the world coordinate system. , , , The maximum speed, acceleration, angular velocity, and angular acceleration of the robot; This indicates the absolute value operation.

[0131] 5. Field of view constraints for mobility aid robots:

[0132]

[0133] in, for The walking robot follows the target at the angle relative to the center line of the field of view. Human-machine distance, It is half the field of view of the robot's vision sensor. , The upper and lower limits of the human-machine distance are respectively.

[0134] 6. Safety corridor constraints for mobility aid robots:

[0135]

[0136] This constraint applies to On the interval, where, This represents the minimum safe distance between the mobile walking robot and the obstacle. Furthermore, regarding the following cost, in this embodiment of the invention, the reference trajectory of the mobile walking robot is directly calculated using a pedestrian motion prediction model based on equation (3), specifically as follows:

[0137]

[0138] in, For a fixed following position defined in the pedestrian coordinate system, It is a two-dimensional rotation matrix. , pedestrians respectively The position and direction of movement at any given moment.

[0139] Considering that pedestrian intentions often change rapidly, the safety corridor at adjacent moments may differ significantly. According to the rolling optimization method, the planning result from the previous moment, which serves as the initial solution for model predictive control, is likely not a feasible solution for the current moment. To address this issue, this invention employs the exterior point method and a quadratic penalty function. For pedestrians with… Inequality constraints of this form can be transformed into the following penalty term:

[0140]

[0141] The above method transforms constraints other than the robot's motion model into corresponding penalty functions, allowing the motion model to be directly substituted to eliminate constraints and transform the problem into an unconstrained one. The algorithm assigns a common weight to each penalty function. And a private weight However, considering the inaccuracy of long-term predictions, the algorithm will adjust the calculation based on the decay rate. The weights of each constraint are gradually reduced as the prediction time increases, focusing on constraints from more recent moments to prevent over-obstacle avoidance. Specifically, the penalty function after constraint transformation takes the following form:

[0142] 7. Exercise range penalty:

[0143]

[0144] in, This indicates the magnitude constraints on the speed (linear velocity) and angular velocity of the mobile walking robot. This indicates the amplitude constraints on the acceleration (linear acceleration) and angular acceleration of the mobile walking robot; , , , These are the weight coefficients for the corresponding items, taken from empirical values; , , , Mobile walking assistance robots The actual velocity, acceleration, angular velocity, and angular acceleration at any given moment.

[0145] 8. Field of view penalty:

[0146]

[0147] In the formula, , These are the weight coefficients corresponding to the field of view penalty function and the human-machine distance penalty function, respectively, and are empirically selected.

[0148] 9. Safety corridor punishment:

[0149]

[0150] In the formula, The weighting coefficients corresponding to the penalty function for the safe corridor are empirically determined.

[0151] The following finite-time optimization control problem can then be obtained:

[0152]

[0153]

[0154] This constraint applies to the initial state and motion model constraints of the mobile assistive robot.

[0155] in, This is the sequence of control variables for the robot. The initial position and initial direction of movement of the walking robot. For the initial velocity and angular velocity of the walking robot, For walking robots Position and direction of movement at any given moment For robots In this embodiment of the invention, velocity refers to linear velocity and acceleration refers to linear acceleration, in order to distinguish between angular velocity and angular acceleration. Here is the state transition matrix. To control the input matrix, This refers to the control variables for the robot.

[0156] The BFGS algorithm is then used to solve the unconstrained optimization problem, thus avoiding the problem of encountering infeasible solutions that prevent further optimization. After optimization, based on the rolling optimization strategy of the MPC algorithm framework, the algorithm is then used... In Control the mobile mobility robot. Substituting the motion model of the mobile walking robot, we obtain the position of the mobile walking robot at the current moment. .

[0157] This invention provides a pedestrian path search algorithm based on pedestrian motion state estimation and obstacle orientation information. Then, by constructing a convex polygon safety area constraint along the search path time by time, a safety corridor can be formed. Furthermore, by constructing corresponding costs and constraints, a pedestrian following and obstacle avoidance method algorithm based on model predictive control and convex polygon safety corridor is proposed to realize pedestrian following, obstacle avoidance and target visibility maintenance in unstructured obstacle environments.

[0158] Example 2

[0159] This invention provides a mobile assistive robot controller for patients with lower limb muscle weakness, including a computer-readable storage medium and a processor;

[0160] Computer-readable storage media are used to store executable instructions;

[0161] The processor is used to read executable instructions stored in a computer-readable storage medium and execute the mobile assistive robot control method for patients with lower limb muscle weakness described in Embodiment 1 above.

[0162] The relevant technical solutions are the same as above, and will not be repeated here.

[0163] Example 3

[0164] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the mobile assistive robot control method for patients with lower limb muscle weakness as described in Embodiment 1 above.

[0165] The relevant technical solutions are the same as above, and will not be repeated here.

[0166] Example 4

[0167] This invention provides an assistive robot for patients with lower limb muscle weakness, comprising: a controller, which executes the control method in Embodiment 1 above, or the controller in Embodiment 2, to perform follow control on the patient with lower limb muscle weakness.

[0168] The relevant technical solutions are the same as above, and will not be repeated here.

[0169] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A mobile walking aid robot control method for a patient with lower extremity muscle weakness, characterized by, The method comprises: predicting a walking path of a target followed by a mobile walking aid robot at N time steps in the future, and controlling the mobile walking aid robot based on the walking path to perform follow-up control on the target; The method further comprises: Step 1: Place the target in The predicted position at any given moment is used as a node in a binary tree, and this node is used as the current node to determine the desired direction of movement in an accessible environment. Is this a feasible direction to travel? If so, then proceed. directional extension distance If the target is not the root node, proceed to step three if the current node is the root node, and proceed to step two if the current node is not the root node. The target's position at the initial moment is the root node of the binary tree. Step two: judging whether the direction extended by the parent node of the current node is a feasible walking direction, if yes, extending the distance along the direction extended by the parent node to reach the child node of the current node, wherein the direction extended by the parent node is the predicted motion direction of the target at the target moment; if not, entering step three Step three: judging whether the direction extended by the parent node of the current node is a feasible walking direction, if yes, extending the distance along the direction extended by the parent node to reach the child node of the current node, wherein the direction extended by the parent node is the predicted motion direction of the target at the target moment; if not, entering step four​ Step three: taking the current expected moving direction as the center to stepwise judge whether there are turnable directions on the left and right sides according to the turn angle from small to large, wherein the turn angle does not exceed a preset first turn angle ; if both sides have turnable directions, a turnable direction with the smallest turn angle is selected to extend a distance to reach a child node of the current node; if only one side has a turnable direction, a turnable direction with the smallest turn angle on the side is selected to extend a distance to reach a child node of the current node; if not, step four is entered ;​​​ Step 4: Move in the desired direction. Based on the center, judge step by step according to the detour angle from small to large. Are there any detour directions to the left and right, where the detour angle is within a preset first detour angle? and the preset second detour angle between, If there are alternative routes on both sides, then extend the distance along the alternative route with the smallest detour angle on both sides. Reach the child nodes of the current node; if there is only one possible detour direction, extend the distance along the detour direction with the smallest detour angle on that side. If not, proceed to step five; Step five: setting a child node of the current node to represent a stop motion state, and the child node has the same extension direction as the current node and an extension distance of 0; Step six: take the child nodes of the current node as the predicted position of the moment, let , enter the iterative search process of step one to step five until N times of iterative search is reached, and a binary tree representing the predicted target walking path is obtained, each branch of the binary tree representing a possible walking path; Also included is constructing The convex polygon safety area at each time point is connected in series on each branch of the binary tree to form a safe corridor without obstacles and obstructions, which is a passable area for the mobile walking robot during the following control process of the target; wherein the constructing The convex polygon safety area at each time point is connected in series on each branch of the binary tree to form a safe corridor without obstacles and obstructions, which is a passable area for the mobile walking robot during the following control process of the target; wherein the constructing Will The predicted target location is used as the center. The center is spaced at equal angular intervals. The plane is divided into Each sector , , Preset experience values; Compute the projection of each obstacle in the first sector to the line connecting the center of the first sector to the center of the second sector and take the minimum :​​​ Among them, the Unit vector of each sector , characterizing the The direction of the angle bisector of each sector Indicates transpose. For the first The angle between the angle bisector of each sector and the x-axis in the world coordinate system. This represents the set of visible obstacles in the world coordinate system. Indicates the first The set of visible obstacles in each sector Indicates the first Any visible obstacle within a sector, Given the distance between the boundary of the safe zone and the center The maximum distance; The resulting The convex polygon security region at the time instant t is represented by the following set: wherein , , ; In step one, the current desired moving direction in the barrier-free environment is determined whether the walking direction is feasible, comprising: If the current desired moving direction The sector into which the obstacle falls Satisfies Then Is a feasible walking direction; wherein, Is the minimum value of the angle bisector projection of the line connecting the moment predicted target position and the sector Each obstacle in the sector To the sector The minimum value of the angle bisector projection, Is a preset empirical coefficient; In step two, determining whether the extension direction of the parent node of the current node is a feasible walking direction comprises: The target is in The sector into which the direction of motion at the time falls Satisfies The direction extended by the parent node of the current node is a feasible walking direction; wherein, The sector The line connecting each obstacle in the sector The minimum value of the angle bisector projection to the sector The predicted target position at the time In step three, whether there is a turnable direction on the left and right sides of the current expected moving direction is judged from small to large according to the turn angle whether there is a turnable direction on the left and right sides of the current expected moving direction By sector Scanning from the center outwards gradually to both sides. Grid sector, , Represents the preset first detour angle; if the first The sector the step falls into satisfy Then in If there is a circumventable direction on the left, end the left-side scan; if the first... The sector the step falls into satisfy Then in There is a possible detour on the right, so the right-side scan ends; among them, , , Representing sectors , Internal obstacles and The minimum value of the projection from the predicted target position line to the corresponding sector angle bisector; the detour direction with the smallest detour angle is the direction of the angle bisector of the sector with the smallest detour angle; In step four, whether there is a turnable direction on the left and right sides of the current position of the vehicle is determined according to the turn angle from small to large, taking the current expected moving direction as the center. the left and right sides of the current position of the vehicle is determined according to the turn angle from small to large, taking the current expected moving direction as the center.​ By sector Centered on, and simultaneously radiating outwards to both sides from the first The grid sector scan reached the first Grid sector, , Characterizes the preset second detour angle; if the first The sector the step falls into satisfy Then in If there is a circumventable direction on the left, end the left-side scan; if the first... The sector the step falls into satisfy Then in There is a possible detour on the right; end the right-side scan. .

2. The mobile walking support robot control method for a lower extremity muscle weakness patient according to claim 1, wherein The method further comprises expanding the convex polygon safety area, comprising: moving the first side of the convex polygonal safety region outward in sequence, the principle of translation being that if the first side and the other sides have a common obstacle to be excluded, then moving the first side to the obstacle corresponding to the minimum of the projections excluded only by the first side; outward in sequence, the principle of translation being that if the first side and the other sides have a common obstacle to be excluded, then moving the first side to the obstacle corresponding to the minimum of the projections excluded only by the first side; outward in sequence, the principle of translation being that if the first side and the other sides have a common obstacle to be excluded, then moving the first side to the obstacle corresponding to the minimum of the projections excluded only by the first side; outward in sequence, the principle of translation being that if the first side and the other sides have a If the first After the obstacle is eliminated by the first edge and other edges, there is no obstacle that is eliminated by the first edge alone. If an obstacle is eliminated by an edge, then the first edge will be eliminated. The strip is shifted outwards to At this time, the first Strip edge distance from center The maximum distance is .

3. The control method of the mobile walking support robot for a lower extremity muscle weakness patient according to claim 1, wherein calculating an approximation degree of each possible walking path in the binary tree to the planned path of the mobile walking aid robot , selecting a walking path corresponding to the minimum as a target most possible walking path, and controlling the mobile walking aid robot based on the target most possible walking path to perform follow-up control on the target; wherein, In the formula, represents the current time, represents the planned path of the mobile walking aid robot at the previous time.

4. The mobile walking support robot control method for a lower extremity muscle weakness patient according to claim 3, wherein controlling the mobile walking aid robot based on the most likely walking path of the target to perform follow-up control on the target, comprising: determining a field of view cost and a field of view constraint of the mobile walking aid robot based on the target position at each time step predicted in the walking path, constructing a safety corridor constraint of the mobile walking aid robot based on the safety corridor corresponding to the walking path, and converting the motion amplitude constraint, the field of view constraint and the safety corridor constraint of the mobile walking aid robot into corresponding quadratic penalty functions; constructing a finite time domain optimal control problem with the sum of the follow-up cost, the obstacle avoidance cost, the field of view cost and the quadratic penalty functions as an objective function, and solving the control variables of the mobile walking aid robot to control the follow-up motion of the mobile walking aid robot under the motion model constraint of the mobile walking aid robot; The follow-up cost is: wherein represents actual and desired position errors, and actual and desired linear velocity errors of the mobile assist robot; represents actual and desired linear acceleration errors of the mobile assist robot; are forward and lateral position error components, respectively, are forward and lateral linear velocity errors, respectively, are forward and lateral linear acceleration error components, respectively; are weights of the corresponding terms.​​​​​​​​ The safety corridor constraint is: wherein is the actual position of the mobile walking aid robot at the moment, is the actual position of the mobile walking aid robot at the moment, is the minimum safety distance of the mobile walking aid robot from the obstacle.

5. The mobile walking support robot control method for a lower extremity muscle weakness patient according to claim 4, wherein The motion amplitude constraint corresponds to a quadratic penalty function: wherein, is the amplitude constraint on the linear and angular velocities of the mobile walking robot, is the amplitude constraint on the linear and angular accelerations of the mobile walking robot; is the preset public weight, is the private weight; , , , is the maximum linear and angular velocities and accelerations of the mobile walking robot; , , , are the actual linear and angular velocities and accelerations of the mobile walking robot at the time instant ; , , , are the weight coefficients of the linear and angular velocities and accelerations; is the absolute value operation; is the two-norm operation; Field of view constraints correspond to a quadratic penalty function is: In the formula, is the angle between the walking aid robot and the target relative to the center line of the field of view, is the distance between the human and the walking aid robot, is half of the field of view angle of the vision sensor of the walking aid robot, , are the upper and lower limits of the distance between the human and the walking aid robot, respectively; , are the weight coefficients corresponding to the field of view angle and the distance between the human and the walking aid robot, respectively. Safety corridor constraints correspond to quadratic penalty functions f : In the formula, is a weight coefficient of the corresponding quadratic penalty function for the safety corridor constraint.

6. A mobile walking aid robot controller for a patient with lower extremity muscle weakness, characterized by, The method comprises a computer readable storage medium and a processor. The computer readable storage medium is configured to store executable instructions. The processor is configured to read the executable instructions stored in the computer readable storage medium to perform the mobile walking aid robot control method for patients with reduced lower limb muscle strength according to any one of claims 1-5.

7. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the mobile walking aid robot control method for patients with reduced lower limb muscle strength according to any one of claims 1-5.

8. A mobile walking support robot for a patient with lower extremity muscle weakness, characterized by, The method comprises a controller configured to perform the mobile walking aid robot control method for patients with reduced lower limb muscle strength according to any one of claims 1-5 to perform follow-up control on the target, or the controller is the mobile walking aid robot controller for patients with reduced lower limb muscle strength according to claim 6.

Images