Robot foot point planning method and device, equipment and storage medium

Abstract

The application discloses a robot landing point planning method and device, equipment and storage medium, relates to the technical field of robot control, and discloses a robot landing point planning method, comprising: acquiring terrain perception information, robot state data, motion accessibility constraint data and collision risk assessment data; generating candidate landing points of the robot based on the terrain perception information and the robot state data; constructing stability indexes, energy consumption indexes and obstacle indexes of the candidate landing points based on the terrain perception information, the robot state data, the motion accessibility constraint data and the collision risk assessment data; determining candidate landing point score information based on the stability indexes, the energy consumption indexes and the obstacle indexes; and determining the target landing point of the robot from the candidate landing points based on the candidate landing point score information to complete the landing point planning of the robot. The scheme effectively improves the passing capacity of the robot in complex terrain and reduces the risk of falling.

Description

Technical Field

[0001] This application relates to the field of robot control technology, and in particular to methods, apparatus, equipment and storage media for robot foot placement planning. Background Technology

[0002] Currently, robot landing point planning technology is mainly divided into two categories: heuristic methods based on geometric rules and path planning methods based on traditional optimization. Heuristic methods based on geometric rules determine the landing area by manually setting explicit selection rules, such as requiring the foot to land on a flat area with minimal elevation difference, or requiring landing points to maintain specific symmetry and spacing. Path planning methods based on traditional optimization, on the other hand, theoretically can generate better landing point schemes by establishing mathematical optimization models and considering certain constraints.

[0003] However, heuristic methods based on geometric rules rely heavily on manually designed rules. While they can maintain a certain level of stability in structured and regular environments, their coverage is limited when facing rugged, dynamic, or unknown complex terrain. They cannot flexibly adapt to terrain changes, easily leading to robots falling or getting stuck due to improper foothold selection. Traditional optimization-based path planning methods, on the other hand, rely on high-precision environmental modeling, resulting in high computational complexity and insufficient real-time performance, making them ill-suited for handling rapid robot movement or sudden environmental changes. Consequently, existing robots exhibit poor stability and environmental adaptability when navigating complex terrain.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a method, apparatus, device and storage medium for planning the landing points of a robot, which aims to solve the technical problem that existing robots have poor stability and environmental adaptability when traversing complex terrain.

[0006] To achieve the above objectives, this application proposes a method for planning the landing points of a robot, the method comprising: Acquire terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; Candidate landing points for the robot are generated based on the terrain perception information and the robot state data; Based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data, the stability index, energy consumption index, and obstacle crossing index of the candidate landing point are constructed. The candidate landing point scoring information is determined based on the stability index, the energy consumption index, and the obstacle crossing index. Based on the candidate landing point scoring information, the target landing point of the robot is determined from the candidate landing points to complete the robot's landing point planning.

[0007] In one embodiment, the step of constructing the stability index, energy consumption index, and obstacle crossing index of the candidate landing point based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data includes: Based on the robot state data, obtain the robot's base posture angle information, center of mass projection information, foot state data, joint state data, and angular velocity; Based on the base attitude angle information, the centroid projection information, and the foot state data, a stability index for the candidate foot landing point is constructed. The energy consumption index of the candidate landing point is constructed based on the joint state data and the angular velocity. The obstacle crossing index of the candidate landing point is constructed based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data.

[0008] In one embodiment, the foot state data includes foot support area information and foot tangential velocity; The step of constructing the stability index of the candidate landing point based on the base attitude angle information, the centroid projection information, and the foot state data includes: The attitude stability reward is determined based on the base attitude angle information; The support margin bonus is determined based on the centroid projection information and the foot support area information. The anti-slip stability bonus is determined based on the foot tangential velocity. The stability index of the candidate foot placement point is constructed based on the posture stability reward, the support margin reward, and the anti-slip stability reward.

[0009] In one embodiment, the joint status data includes the number of joints, current joint torque, historical joint torque, and torque acquisition time interval; The step of constructing the energy consumption index of the candidate landing point based on the joint state data and the angular velocity includes: The power consumption reward is determined based on the current joint torque, the angular velocity, and the number of joints. A torque smoothing reward is determined based on the current joint torque, the historical joint torque, and the torque acquisition time interval. The energy consumption index of the candidate landing point is constructed based on the power energy consumption reward and the torque smoothing reward.

[0010] In one embodiment, the terrain perception information includes obstacle height information, the height variance, slope angle, and normal of the candidate landing area, and the motion accessibility constraint data includes the robot's leg swing height and leg reachability range. The step of constructing the obstacle-crossing index of the candidate landing point based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data includes: The leg swing space reward is determined based on the obstacle height information and the leg swing height. The terrain support reward is determined based on the height variance, the slope angle, and the normal. No-collision bonuses are determined based on collision risk assessment data. The leg reachability reward is determined based on the coordinates of the candidate foot placement point and the leg reachability range. The obstacle-crossing index of the candidate landing point is constructed based on the leg swing space reward, the terrain support reward, the collision-free reward, and the leg reachability reward.

[0011] In one embodiment, the step of generating candidate landing points for the robot based on the terrain perception information and the robot state data includes: Based on the robot state data, robot gait parameters and robot posture prediction information are obtained; Candidate landing points for the robot are generated based on the terrain perception data, the robot gait parameters, and the robot posture prediction information.

[0012] In one embodiment, the step of generating candidate footholds for the robot based on the terrain perception data, the robot gait parameters, and the robot posture prediction information includes: The target swing leg and desired foot placement position of the robot are determined based on the robot's gait parameters. Based on the robot posture prediction information and the target swing leg, a physically reachable search area centered on the desired foot landing point is determined. Sampling is performed within the physically reachable search area to obtain multiple candidate landing points.

[0013] Furthermore, to achieve the above objectives, this application also proposes a robot foot placement planning device, which includes: The data acquisition module is used to acquire terrain perception information, robot status data, motion accessibility constraint data, and collision risk assessment data. The data processing module is used to generate candidate landing points for the robot based on the terrain perception information and the robot state data; The data processing module is also used to construct stability indicators, energy consumption indicators, and obstacle crossing indicators for the candidate landing points based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data. The landing point scoring module is used to determine candidate landing point scoring information based on the stability index, the energy consumption index, and the obstacle crossing index. The landing point determination module is used to determine the target landing point of the robot from the candidate landing points based on the candidate landing point scoring information, so as to complete the landing point planning of the robot.

[0014] In addition, to achieve the above objectives, this application also proposes a robot footing planning device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the robot footing planning method as described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the robot foot placement planning method described above.

[0016] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the robot foot placement planning method described above.

[0017] One or more technical solutions proposed in this application have at least the following technical effects: By comprehensively acquiring terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data, candidate landing points are generated based on this data. This ensures the physical accessibility of the landing points while avoiding ineffective areas based on actual terrain conditions. Three key indicators—stability, energy consumption, and obstacle clearance—are constructed to comprehensively evaluate candidate landing points from different core dimensions. The scores calculated based on these three indicators objectively reflect the quality of the landing points. The final selected target landing points achieve the optimal combination of stability, energy efficiency, and terrain adaptability. This approach effectively improves the robot's ability to navigate complex terrain, reduces the risk of falls, and is flexibly applicable to various mobile platforms, providing an efficient and intelligent landing point planning solution for autonomous movement and operations in multiple scenarios. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an embodiment of the robot's landing point planning method in this application. Figure 2 This is a flowchart illustrating the second embodiment of the robot's landing point planning method in this application. Figure 3 This is a schematic diagram of the module structure of the robot's foot placement planning device according to an embodiment of this application; Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the robot's landing point planning method in the embodiments of this application.

[0021] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0023] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The main solution of this application embodiment is as follows: acquiring terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; generating candidate landing points for the robot based on terrain perception information and robot state data; constructing stability indicators, energy consumption indicators, and obstacle crossing indicators for the candidate landing points based on terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; determining the candidate landing point scoring information based on the stability indicators, energy consumption indicators, and obstacle crossing indicators; and determining the robot's target landing point from the candidate landing points based on the candidate landing point scoring information to complete the robot's landing point planning.

[0025] Currently, robot landing point planning technology is mainly divided into two categories: heuristic methods based on geometric rules and path planning methods based on traditional optimization. Heuristic methods based on geometric rules determine the landing area by manually setting explicit selection rules, such as requiring the foot to land on a flat area with minimal elevation difference, or requiring landing points to maintain specific symmetry and spacing. Path planning methods based on traditional optimization, on the other hand, theoretically can generate better landing point schemes by establishing mathematical optimization models and considering certain constraints.

[0026] However, heuristic methods based on geometric rules rely heavily on manually designed rules. While they can maintain a certain level of stability in structured and regular environments, their coverage is limited when facing rugged, dynamic, or unknown complex terrain. They cannot flexibly adapt to terrain changes, easily leading to robots falling or getting stuck due to improper foothold selection. Traditional optimization-based path planning methods, on the other hand, rely on high-precision environmental modeling, resulting in high computational complexity and insufficient real-time performance, making them ill-suited for handling rapid robot movement or sudden environmental changes. Consequently, existing robots exhibit poor stability and environmental adaptability when navigating complex terrain.

[0027] This application provides a solution that comprehensively acquires terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data. Based on this data, candidate landing points are generated, ensuring both the physical accessibility of the landing points and avoiding ineffective areas by considering the actual terrain conditions. Three key indicators—stability, energy consumption, and obstacle clearance—are constructed to comprehensively evaluate candidate landing points from different core dimensions. The scores calculated based on these three indicators objectively reflect the quality of the landing points. The final selected target landing points achieve the optimal combination of stability, energy efficiency, and terrain adaptability. This approach effectively improves the robot's ability to navigate complex terrain, reduces the risk of falls, and is flexibly applicable to various mobile platforms, providing an efficient and intelligent landing point planning solution for autonomous movement and operations in multiple scenarios.

[0028] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a robot's landing point planning device. The following description uses a robot's landing point planning device as an example to illustrate this embodiment and the subsequent embodiments.

[0029] Based on this, embodiments of this application provide a method for planning the landing points of a robot, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the robot's landing point planning method according to this application.

[0030] In this embodiment, the robot's landing point planning method includes steps S10 to S50: Step S10: Acquire terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; It should be noted that terrain perception information, which is data related to the terrain around the robot collected by sensors, including the actual height of steps or other obstacles, the slope of the terrain, height difference, surface normal direction, and geometric features such as support area, is the key basis for judging the terrain adaptability of the landing point.

[0031] In addition, robot state data describes the robot's current motion state and body posture, including the roll and pitch angles of the robot base, data related to the projection of the robot's center of mass onto the support surface, data related to the movement and contact state of the robot's feet, the attributes and motion-related data of the robot's joints, the torque data of each joint at the current moment, the torque data of each joint at the previous moment, the torque acquisition time interval, the angular velocity data of the robot's overall motion, and historical gait sequences, which can reflect the robot's current motion capabilities and posture.

[0032] In addition, motion accessibility constraint data is relevant data that limits the robot's leg swinging motion and foot landing range. This includes data on the maximum height the robot's swinging leg can lift, data on the spatial range that the foot can reach based on parameters such as the robot's joint kinematic range and leg length limitations, etc., which provide physical boundary constraints for the generation of candidate foot landing points.

[0033] In addition, collision risk assessment data is used to determine whether there is a potential collision hazard at the candidate landing point. It is calculated by combining terrain obstacle information and robot model, and directly affects the safety of the landing point.

[0034] It should be understood that the surrounding terrain data is collected by sensors such as LiDAR, depth cameras or height map scanning modules to obtain terrain perception information, the robot's own sensors detect joint forces, movement speed and other data to obtain robot state data, the robot's structural parameters and joint performance parameters are combined to determine motion accessibility constraint data, and the relationship between the terrain and the robot model is analyzed by collision detection algorithms to obtain collision risk assessment data.

[0035] Step S20: Generate candidate landing points for the robot based on the terrain perception information and the robot state data; It should be noted that candidate landing points are a series of discrete or continuous locations within the physically accessible space of the robot's legs, selected based on surrounding terrain features and the robot's own motion state. These locations are filtered from the physically accessible search area to provide a range of choices for further selection of target landing points. The robot can be a quadruped, wheeled, or humanoid robot, among other mobile platforms.

[0036] It should be understood that when generating candidate landing points, the nominal landing point in the ideal state is calculated based on the gait parameters in the robot's state data, such as stride length, stride frequency, and gait phase. Then, based on the robot's current body posture, such as center of mass position, pitch angle, and leg length limitations, and combined with terrain perception information, the reachable search area is determined with the nominal landing point as the center. Within this area, candidate landing points for the robot are generated through methods such as grid sampling and random sampling.

[0037] Step S30: Based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data, construct the stability index, energy consumption index, and obstacle crossing index of the candidate landing point; It should be noted that the stability index is a core indicator that comprehensively measures the robot's posture stability, support stability, and anti-slip performance during gait execution when using the candidate foothold. It can fully reflect whether the robot will experience instability such as tipping or slipping due to improper foothold selection, and is a key indicator to ensure the robot's safe movement.

[0038] In addition, energy consumption index is an important indicator used to evaluate the energy efficiency and joint torque smoothness of the robot when it moves with the candidate foot point. The higher the energy efficiency and the smoother the torque change, the better the energy consumption index, which can effectively reduce the energy consumption of the robot during the movement process, reduce the wear and tear of mechanical structures such as joints, and extend the working time of the robot.

[0039] In addition, obstacle crossing index is a core indicator for evaluating the accessibility, terrain support and collision-free nature of a candidate landing point when a robot crosses obstacles or terrain with elevation differences. It is directly related to whether the robot can successfully cross the current obstacle, adapt to complex terrain with elevation differences and improve the robot's ability to pass through complex environments.

[0040] It should be understood that when constructing stability indices, posture stability-related parameters are calculated by combining the base posture angle information from the robot's state data, support margin-related parameters are calculated based on the centroid projection information and foot support area information, and anti-slip performance-related parameters are calculated using foot tangential velocity data. Through the comprehensive integration of these parameters, an index that can comprehensively reflect the stability of the robot when using the candidate footing point is constructed. When constructing energy consumption indices, the mechanical power and torque change rate of each joint are calculated based on the number of joints, current joint torque, historical joint torque, torque acquisition time interval, and angular velocity from the robot's state data. Based on this, power-related energy consumption parameters and torque smoothing penalty term-related parameters are constructed, and these two types of parameters are integrated to form a complete energy consumption index. When constructing obstacle crossing indicators, the clearance-related parameters are calculated by combining obstacle height information from terrain perception information and leg swing height from motion accessibility constraint data. Terrain support-related parameters are constructed based on the height variance, slope angle, and normal of the candidate landing area. Collision risk assessment data is used to determine collision-free related parameters. Accessibility-related parameters are calculated by combining the leg accessibility range from motion accessibility constraint data. These parameters are systematically integrated to form obstacle crossing indicators that can comprehensively evaluate obstacle crossing ability.

[0041] In one feasible implementation, step S30 may include steps S31 to S34: Step S31: Based on the robot state data, obtain the robot's base posture angle information, center of mass projection information, foot state data, joint state data, and angular velocity; It should be noted that the base attitude angle information is data describing the robot base attitude, including roll and pitch angles, used to measure the attitude stability of the robot base. The center of mass projection information is data related to the projection point of the robot's center of mass onto the support surface, including the position of this projection point relative to the support polygon, and is the basis for evaluating support stability.

[0042] Additionally, foot state data is crucial data related to the robot's foot movement and contact state. This includes the range and shape of the contact area between the foot and the ground, as well as the foot's tangential velocity along the ground. This data directly reflects the foot's contact with the ground and its movement state, used to assess the foot's anti-slip performance. Joint state data comprises the robot's joint attributes and motion-related data, including the number of joints, current joint torque, historical joint torque, and torque acquisition time intervals. Joints are the core actuators of robot motion, and joint state data comprehensively reflects the joint's load, motion changes, and smoothness. Angular velocity is the instantaneous speed of the robot's overall motion, reflecting the speed of joint movement. Its magnitude directly affects the joint's work efficiency and motion stability.

[0043] It should be understood that by classifying, extracting, and parsing the robot's state data, the raw data related to the base posture is separated from the data, and the base posture angle information is obtained after calculation and processing; parameters related to the center of mass position are filtered from the robot's state data, and the center of mass projection information is derived by combining the spatial position relationship of the support surface; records related to the foot end are extracted from the robot's state data, including the description of the contact area between the foot end and the ground and the tangential component data of the foot end movement velocity, and integrated to form foot end state data; structural parameters and motion data related to the joints are separated, and the number of joints, the current and historical joint torque values, and the time interval of torque acquisition are identified to constitute joint state data; at the same time, velocity-related data of joint or overall motion are extracted from the robot's state data, and angular velocity is obtained through orientation analysis.

[0044] Step S32: Construct a stability index for the candidate landing point based on the base attitude angle information, the centroid projection information, and the foot state data; It should be understood that the attitude stability-related parameters are calculated using the base attitude angle information, the support margin is evaluated by combining the centroid projection information with the support polygon data, and the anti-slip performance is judged based on the foot tangential velocity in the foot state data. The evaluation results of these three aspects are integrated to construct a stability index that comprehensively reflects the stability performance of the candidate landing point.

[0045] In one feasible implementation, the foot state data includes foot support area information and foot tangential velocity; step S32 may include steps S321~S324: Step S321: Determine the attitude stability reward based on the base attitude angle information; It should be noted that the attitude stability reward is a pre-set reward value used to measure the attitude stability of the robot base. Its value ranges from 0 to 1, and the higher the value, the more stable the attitude of the base.

[0046] It should be understood that the roll angle and pitch angle in the base attitude angle information are obtained, and the deviation of these two angles is converted into a value in the range of 0 to 1 through the definition formula of attitude stability reward. This value is the attitude stability reward, which is used to quantify the stability of the base attitude.

[0047] For example, the formula for defining the attitude stability reward is as follows:

[0048] In the formula, This represents the attitude stability reward value, used to measure the stability of the robot's base attitude; This represents an exponential function used to convert attitude angle deviations into reward values ​​in the 0-1 range; Indicates the roll angle of the base; Indicates the pitch angle of the base; This represents the pre-set attitude angle standard deviation parameter used to adjust the degree of influence of attitude angle deviation on the reward.

[0049] Alternatively, it can be based on the projection of gravity in the body coordinate system. The attitude stability reward is determined using the following formula, which calculates the stability reward based on the projected attitude of gravity in the body coordinate system:

[0050] In the formula, This represents the projection vector of gravity in the robot's body coordinate system, used to characterize the component distribution of gravity in the robot's own coordinate system; This represents the projected component of gravity along the x-axis of the body coordinate system, which is calculated by the dot product of the gravity vector and the unit vector along the x-axis of the body coordinate system. This represents the projected component of gravity along the y-axis of the body coordinate system, calculated by the dot product of the gravity vector and the unit vector of the y-axis of the body coordinate system. This represents the projected component of gravity along the z-axis of the system, which is calculated by the dot product of the gravity vector and the unit vector along the z-axis of the system. This represents the pre-set gravity standard deviation parameter used to adjust the degree of influence of the gravity plane component on the reward.

[0051] Step S322: Determine the support margin bonus based on the centroid projection information and the foot support area information; It should be noted that the foot support area information reflects the data of the contact area between the robot's foot and the ground, including the shape and extent of the support area. Based on this information, the support polygon formed by the robot's current contact foot can be determined. The support margin reward is a pre-set reward value used to evaluate the stability margin of the robot's centroid projection point relative to the support polygon. The value ranges from 0 to 1, with a higher value indicating a greater stability margin.

[0052] It should be understood that, by combining the centroid projection point in the centroid projection information and the foot support area information to form a support polygon, the distance from the centroid projection point to the boundary of the support polygon is calculated. This distance is then compared with a preset safety distance threshold and converted into a support margin reward in the range of 0 to 1, thereby assessing the support stability.

[0053] For example, the formula for defining the margin reward is as follows:

[0054] In the formula, This represents the support margin reward value, used to evaluate the stability margin of the robot's centroid projection point relative to the supporting polygon. This represents a cutoff function used to restrict the calculation result to the interval between 0 and 1; Used to calculate the distance from the centroid projection point to the boundary of the supporting polygon. This represents the projection point of the robot's center of mass. This represents the supporting polygon formed by the robot's current contact foot. This represents a pre-set safe distance threshold used to determine whether the centroid projection point is within a safe range.

[0055] Step S323: Determine the anti-slip stability bonus based on the foot tangential velocity; It should be noted that the foot tangential velocity is the speed at which the robot's foot moves along the tangent to the ground. The magnitude of this velocity is related to whether the foot will slip; the higher the velocity, the higher the risk of slipping. The anti-slip stability bonus is a pre-set bonus value used to measure the anti-slip performance of the foot, ranging from 0 to 1. The higher the value, the better the anti-slip performance and the lower the risk of slipping.

[0056] It should be understood that the number of feet currently in contact with the ground is counted, the tangential velocity of each contacting foot is obtained and its modulus is calculated, and the tangential velocity modulus is converted into a slip-related value for each foot using the definition formula of slip stability reward. Then, the average of this value for all contacting feet is calculated to obtain the slip stability reward.

[0057] For example, the formula for defining the anti-slip stability bonus is as follows:

[0058] In the formula, This represents the anti-slip stability bonus value, used to measure the anti-slip performance of the foot. The constant representing the number of feet is a pre-set parameter used to normalize the foot slip resistance bonus; This represents the sum of all foot anti-slip rewards. Indicates the single foot tip currently in contact with the ground; Indicates the first The tangential velocity of each foot is calculated by detecting the foot's motion state using sensors; Indicates the first The modulus of the tangential velocity at the foot tip; This represents the pre-set speed standard deviation parameter used to adjust the degree of influence of foot tangential speed on anti-slip reward.

[0059] Step S324: Construct a stability index for the candidate landing point based on the attitude stability reward, the support margin reward, and the anti-slip stability reward.

[0060] It should be noted that the stability index is the core indicator that comprehensively measures the stability performance of the candidate footing point. It integrates the evaluation results of three key dimensions: posture stability, support stability, and anti-slip stability, and comprehensively reflects the stability of the robot when performing gait at that footing point.

[0061] It should be understood that when constructing stability metrics, corresponding weight parameters are set for posture stability reward, support margin reward, and anti-slip stability reward based on the robot's motion scenario and requirements. The weight parameters are set specifically; for example, in rugged terrain, posture stability and support stability are particularly important, so the weights of posture stability reward and support margin reward are appropriately increased; in smooth terrain, anti-slip stability is more critical, so the weight of anti-slip stability reward is appropriately increased. Then, the posture stability reward is multiplied by its corresponding weight parameter to obtain the weighted posture stability reward; the support margin reward is multiplied by its corresponding weight parameter to obtain the weighted support margin reward; and the anti-slip stability reward is multiplied by its corresponding weight parameter to obtain the weighted anti-slip stability reward. The sum of these three weighted reward values ​​is then calculated, and the total sum is the stability metric for the candidate landing point.

[0062] For example, the formula for defining the stability index is as follows:

[0063] In the formula, The value of the stability index is used to comprehensively evaluate the robot's posture stability, support stability, and anti-slip performance during gait execution. This represents the pre-set parameters used to adjust the attitude stability reward weights; This represents the attitude stability reward value; Parameters representing the weight of rewards related to gravity projection; This represents the attitude stability bonus calculated based on the projection of gravity onto the body coordinate system. This refers to the pre-set parameters used to adjust the support margin reward weights; This represents the support margin reward value; This represents a pre-set parameter used to adjust the weighting of the anti-slip stability bonus; This indicates the bonus value for anti-slip stability.

[0064] In this embodiment, a posture stability reward is obtained by quantifying posture deviation, which accurately reflects whether the base is tilted. Combining the centroid projection and foot support area information, the stability performance is evaluated from the perspective of support margin, ensuring that the robot's center of gravity is within a safe range. For the key risk point of foot anti-slip, the anti-slip performance is quantified by tangential velocity to avoid instability caused by slippage. By weighted integration of the three sub-rewards, a stability index is constructed, which provides clear data support and objective evaluation standards for the stability index, providing an accurate and reliable basis for the comprehensive scoring of subsequent candidate foot placement points.

[0065] Step S33: Construct the energy consumption index of the candidate landing point based on the joint state data and the angular velocity; It should be understood that the mechanical power of each joint is calculated by using the joint torque and angular velocity in the joint state data, thereby obtaining the overall power consumption related assessment results. At the same time, the torque change rate is calculated by using the current torque and the torque at the previous moment in the joint state data, combined with the time interval and the maximum allowable torque of the joint, to assess the torque smoothness. These two parts are integrated to construct the energy consumption index.

[0066] In one feasible implementation, the joint state data includes the number of joints, current joint torque, historical joint torque, and torque acquisition time interval; step S33 may include steps S331~S333: Step S331: Determine the power consumption reward based on the current joint torque, the angular velocity, and the number of joints; It should be noted that the current joint torque describes the force state of each joint of the robot at the current moment, directly reflecting the load situation during joint movement. The number of joints is the total number of joints in the robot, used to uniformly calculate the overall energy consumption level. The power consumption reward is a pre-set reward value used to measure the robot's motion energy efficiency, ranging from 0 to 1, with higher values ​​indicating higher energy utilization efficiency.

[0067] It should be understood that the mechanical power of each joint is calculated based on the current joint torque and corresponding angular velocity of each joint. The absolute value of the mechanical power of all joints is taken and then averaged. This average value is then compared with the preset reference power value. The power energy consumption reward is converted into a power energy consumption reward in the range of 0 to 1 through the definition formula of power energy consumption reward, thereby quantifying energy efficiency.

[0068] For example, the formula for defining power consumption reward is as follows:

[0069] In the formula, This represents the power consumption reward value, used to measure the energy efficiency of robot movement; This indicates the total number of joints in the robot; Indicates all of the robot Summation operation is performed on each joint; Indicates the first The torque of each joint; Indicates the first Angular velocity of each joint; Indicates the first The absolute value of the mechanical power of each joint; This indicates a pre-set reference power value used as a power reference standard.

[0070] The formula for calculating mechanical power is as follows:

[0071] In the formula, Indicates the first The mechanical power of a joint is used to measure the energy consumption of that joint's movement.

[0072] Step S332: Determine the torque smoothing reward based on the current joint torque, the historical joint torque, and the torque acquisition time interval; It should be noted that historical joint torque is the torque data of each joint of the robot at a previous moment, obtained by recording historical motion data. Combining it with the current joint torque reflects the trend of torque change. The torque acquisition interval is a pre-set time step parameter between two joint torque acquisitions, used to calculate the rate of torque change. The torque smoothing reward is a pre-set reward value to encourage smooth changes in joint torque, ranging from 0 to 1. A higher value indicates a more stable torque change and lower mechanical wear.

[0073] It should be understood that the difference between the current joint torque and the historical joint torque for each joint is calculated. This difference is divided by the torque acquisition time interval to obtain the torque change rate. Then, combined with the maximum allowable torque for each joint, the deviation of the torque change rate relative to the maximum allowable change rate is calculated. This deviation is converted into a value in the range of 0 to 1 through specific calculations. The torque smoothing reward is obtained by averaging this value for all joints.

[0074] For example, the formula for defining the torque smoothing reward is as follows:

[0075] In the formula, This represents the torque smoothing bonus value, used to encourage smooth changes in joint torque and reduce sudden torque changes; This indicates the total number of joints in the robot; Indicates all of the robot Summation operation is performed on each joint; Indicates the first The torque of each joint at the current moment; Indicates the first The torque at a given moment on a joint; This represents the time interval between the current moment and the previous moment, i.e., the torque acquisition time interval; Indicates the first The maximum allowable torque of each joint is a pre-set joint structure performance parameter; Indicates the first The rate of change of joint torque relative to the square of the maximum permissible rate of change.

[0076] Step S333: Construct the energy consumption index of the candidate landing point based on the power energy consumption reward and the torque smoothing reward.

[0077] It should be understood that when constructing energy consumption indicators, firstly, based on the robot's motion requirements and application scenarios, corresponding weight parameters are set for power consumption rewards and torque smoothing rewards. The magnitude of the weight parameters reflects the importance of the two rewards in energy consumption assessment and can be flexibly adjusted according to actual conditions. For example, during long-term operation, energy efficiency is emphasized, and the weight corresponding to the power consumption reward is increased; when moving on complex terrain, torque stability is emphasized. Subsequently, the power consumption reward is multiplied by the corresponding weight parameter to obtain the weighted power consumption reward; the torque smoothing reward is multiplied by the corresponding weight parameter to obtain the weighted torque smoothing reward. The sum of these two weighted reward values ​​is calculated, and the total sum is the energy consumption indicator for the candidate landing point. This indicator can comprehensively and objectively reflect the overall energy consumption-related performance of the robot when moving at the candidate landing point.

[0078] For example, the formula for defining the energy consumption index is as follows:

[0079] In the formula, This represents the value of the energy consumption index, used to comprehensively evaluate the energy efficiency and torque smoothness of robot motion; This refers to the pre-set parameters used to adjust power consumption rewards; This represents the power consumption bonus value; This represents the pre-set parameters used to adjust the torque smoothing reward weights; This represents the torque smoothing bonus value.

[0080] In this embodiment, energy efficiency is precisely quantified by the current joint torque, angular velocity, and number of joints. The resulting power energy consumption reward objectively reflects the energy-saving level of the motion. The torque smoothing reward is calculated based on torque change data, which effectively avoids mechanical losses caused by sudden torque changes and protects the robot's joint structure. By weighted integration of the two rewards, an energy consumption index is constructed, which takes into account both energy utilization efficiency and the stability of mechanical motion, making the energy consumption assessment more comprehensive and scientific. This provides an objective and accurate energy consumption evaluation basis for candidate landing points, which helps to screen out the optimal landing point with low energy consumption and low loss, thereby improving the robot's endurance and mechanical lifespan.

[0081] Step S34: Construct obstacle crossing indicators for the candidate landing points based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data.

[0082] It should be understood that the obstacle crossing index is constructed by using obstacle height, terrain slope, height variance and other data in terrain perception information to assess leg swing clearance and terrain support, combining leg joint kinematic range, leg length restriction and other data in motion accessibility constraint data to determine the accessibility of the landing point, and determining whether there is a collision risk for the candidate landing point based on collision risk assessment data.

[0083] In one feasible implementation, the terrain perception information includes obstacle height information, the height variance, slope angle, and normal of the candidate landing area; the motion accessibility constraint data includes the robot's leg swing height and leg reachability range; step S34 may include steps S341~S345: Step S341: Determine the leg swing space reward based on the obstacle height information and the leg swing height; It should be noted that obstacle height information is specific data reflecting the height of obstacles or steps in terrain perception information. This data is obtained by detecting terrain obstacles through sensors and directly determines the space margin required for the robot to swing its leg over them. Leg swing height is the pre-set maximum leg swing height in the motion accessibility constraint data and is a core parameter of the robot's leg swing capability. Leg swing space reward is a pre-set reward value used to evaluate whether there is sufficient space for the robot to swing its leg over obstacles. The value ranges from 0 to 1; a higher value indicates more leg swing space and higher safety when crossing obstacles.

[0084] It should be understood that by combining the obstacle height information and the preset leg swing height, the spatial difference between the two is calculated. If the obstacle height plus the reserved safety margin parameter does not exceed the leg swing height, then the space is sufficient. Through specific calculations, the degree of space sufficiency is converted into a leg swing space reward in the range of 0 to 1, thereby quantifying the spatial conditions for leg swing crossing.

[0085] For example, the formula for defining leg space reward is as follows:

[0086] In the formula, This represents the leg swing space reward value, used to evaluate whether there is enough space for the robot to swing its legs to cross obstacles; This represents the maximum value function, used to select the larger value between 0 and the insufficient space, ensuring that the penalty only takes effect when there is insufficient headroom; Indicates the height of a step or obstacle; This indicates a pre-set safety margin parameter used to reserve additional headroom; This indicates the maximum height the robot can swing its legs; This represents the pre-set net air standard deviation parameter used to adjust the degree of impact of insufficient net air on rewards.

[0087] Step S342: Determine the terrain support reward based on the height variance, the slope angle, and the normal. It should be noted that height variance is the data on the difference in terrain height within the candidate landing area, calculated from terrain height data collected by sensors, reflecting the flatness of the terrain. Slope angle is the tilt angle of the candidate landing area, calculated from terrain data collected by sensors, affecting stability after landing. Normal is a vector perpendicular to the ground of the candidate landing area, calculated from terrain point cloud data collected by sensors, used to measure the verticality of the terrain surface. Terrain support bonus is a pre-set bonus value used to evaluate the terrain support performance of the candidate landing area, ranging from 0 to 1, with higher values ​​indicating better terrain support.

[0088] It should be understood that the height variance, slope angle, and normal are quantitatively evaluated separately. The height variance is compared with the preset standard height variance, the slope angle is compared with the preset slope angle reference value, and the normal is measured by the dot product with the vertically upward unit vector to measure the degree of verticality. Then, weights are assigned to these three evaluation results, and the weighted sum is converted into a terrain support reward in the range of 0 to 1, which comprehensively reflects the support capacity of the terrain.

[0089] For example, the formula for defining terrain support reward is as follows:

[0090] In the formula, This represents the terrain support bonus value, used to evaluate the terrain support performance of the candidate landing point area; This represents a pre-set parameter used to adjust the reward weights related to height variance; The height variance of the candidate landing point area is calculated using terrain height data collected by sensors. This represents the pre-defined standard height variance parameter used as a reference for height variance. This represents the pre-set parameters used to adjust the reward weights related to the slope angle; Indicates the slope angle of the candidate landing point area; This represents the pre-set slope angle standard deviation parameter used to adjust the degree of influence of slope angle on rewards; This represents a pre-set parameter used to adjust the weights of the normal-related rewards; The surface normal vector representing the candidate landing point area is calculated from terrain point cloud data collected by sensors. This represents the vertical upward unit vector of the reference coordinate system, which is a pre-defined coordinate system reference parameter; It represents the dot product of the surface normal vector and the vertically upward unit vector, and is used to measure the verticality of the terrain surface; This represents the computational term that normalizes the result of the normal dot product to the 0-1 interval.

[0091] Step S343: Determine the no-collision reward based on the collision risk assessment data; It should be noted that the collision risk assessment data is used to determine whether there is a potential collision hazard at the candidate landing point. It is calculated by combining the collision detection algorithm with the terrain and robot model, and is a key basis for assessing the safety of the landing point. The no-collision reward is a pre-set reward value used to assess whether there is a collision risk at the candidate landing point. The value is 0 or 1, with 1 for no collision risk and 0 for a collision risk.

[0092] It should be understood that by analyzing collision risk assessment data, it is determined whether the candidate landing point will collide with terrain obstacles or its own structure. If there is no collision risk, the no-collision reward is set to 1, and if there is a collision risk, the reward is set to 0, thus intuitively reflecting the collision safety of the landing point.

[0093] For example, the formula for defining the collision-free reward is as follows:

[0094] In the formula, This represents the no-collision bonus value, used to assess whether there is a collision risk at the candidate landing point; This represents the collision indication function, which takes the value of 1 if there is a collision risk at the candidate landing point and takes the value of 0 if there is no collision risk. It is calculated by combining the collision detection algorithm with the terrain and robot model.

[0095] Step S344: Determine the leg reachability reward based on the candidate foot landing point coordinates and the leg reachability range; It should be noted that the candidate foothold coordinates are the specific spatial coordinates of each candidate foothold, and are the core information for locating the foothold. The leg reachability range is the spatial range that the robot's legs can reach, pre-calculated based on parameters such as the robot's joint kinematic range and leg length, within the motion reachability constraint data. The leg reachability reward is a pre-set reward value used to evaluate whether a candidate foothold is within the robot's leg reachability range, ranging from 0 to 1, with higher values ​​indicating better reachability.

[0096] It should be understood that the distance from the candidate landing point's coordinates to the leg's reachable domain is calculated, and this distance is converted into a value between 0 and 1 through specific calculations. The closer the distance, the higher the reward value, and the farther the distance, the lower the reward value. In this way, the leg reachability reward is obtained, and the physical reachability of the landing point is quantified.

[0097] For example, the formula for defining the leg reach reward is as follows:

[0098] In the formula, This represents the leg reachability reward value, used to evaluate whether the candidate footing point is within the reachability range of the robot's legs; This represents a distance calculation function used to calculate the distance from a candidate foot landing point to the reachable domain of the leg; Indicates the position coordinates of the candidate landing point; This indicates the reachable range of the robot's legs; This represents the square of the reachable range from the candidate's foot landing point to their leg. Specifically, it indicates the landing point. To reach the region The minimum Euclidean distance between any point within the interior; This represents the standard deviation parameter of the reachability domain, which is preset to adjust the degree of influence of reachability distance on rewards.

[0099] Step S345: Construct obstacle crossing indicators for the candidate landing points based on the leg swing space reward, the terrain support reward, the collision-free reward, and the leg reachability reward.

[0100] It should be understood that when constructing obstacle-crossing metrics, firstly, based on the robot's obstacle-crossing needs and actual movement scenarios, corresponding weight parameters are set for leg-swinging space rewards, terrain support rewards, collision-free rewards, and leg-accessible rewards. These weight parameters are set according to the actual scenario. For example, when crossing higher obstacles, the weights of leg-swinging space rewards and leg-accessible rewards are appropriately increased to emphasize the feasibility of the obstacle-crossing action; in rugged terrain environments, the weight of terrain support rewards is appropriately increased to emphasize the stability after landing; in obstacle-dense scenarios, the weight of collision-free rewards is appropriately increased to emphasize the safety of the landing point. Subsequently, the leg-swinging space reward is multiplied by its corresponding weight parameter to obtain the weighted leg-swinging space reward; the terrain support reward is multiplied by its corresponding weight parameter to obtain the weighted terrain support reward; the collision-free reward is multiplied by its corresponding weight parameter to obtain the weighted collision-free reward; and the leg-accessible reward is multiplied by its corresponding weight parameter to obtain the weighted leg-accessible reward. The sum of these four weighted reward values ​​is the obstacle clearance index of the candidate landing point. This index can comprehensively and accurately reflect the overall performance of the candidate landing point in obstacle clearance.

[0101] For example, the obstacle clearance index is defined by the following formula:

[0102] In the formula, The value of the obstacle crossing index is used to comprehensively evaluate the accessibility and safety of the landing point when the robot crosses obstacles or terrain with elevation differences; This refers to the pre-set parameters used to adjust the weighting of leg swing space rewards; This represents the bonus value for leg swing space; This represents a pre-set parameter used to adjust the terrain support reward weight; This indicates the terrain support bonus value; This represents the pre-set parameters used to adjust the weight of the collision-free reward. This indicates the no-collision bonus value; This refers to the pre-set parameters used to adjust the weight of leg reach rewards; This indicates that the leg area can reach the reward value.

[0103] In this implementation, the leg swing space reward is quantified by comparing the obstacle height with the leg swing height to ensure that no collision occurs when crossing obstacles. A terrain support reward is constructed from three dimensions: terrain flatness, tilt angle, and verticality to comprehensively evaluate the terrain adaptability of the landing point. The collision risk of the landing point is intuitively judged by the collision-free reward. The leg reachability reward is obtained by calculating the distance between the candidate point and the reachable domain to ensure the physical executability of the landing point. This covers the key aspects of the obstacle crossing process and adapts to the needs of different terrain scenarios. It provides a scientific basis for the obstacle crossing performance of the candidate landing point and effectively improves the robot's obstacle crossing ability and passage safety in complex terrain.

[0104] Step S40: Determine candidate landing point scoring information based on the stability index, the energy consumption index, and the obstacle crossing index; It should be noted that the candidate landing point score is a comprehensive result obtained after quantitatively evaluating the stability, energy efficiency, and obstacle-crossing ability of each candidate landing point. Presented in specific numerical form, it clearly and intuitively reflects the differences in quality between different candidate landing points. This score is calculated by assigning corresponding weights to each indicator based on the core requirements of robot movement and actual application scenarios. The weight parameters can be adjusted in a course-based manner according to different training stages or the needs of actual movement tasks to highlight the core requirements in different scenarios.

[0105] It should be understood that corresponding weight parameters are set for stability indicators, energy consumption indicators, and obstacle crossing indicators. These weights can be adjusted in a course-based manner according to the training phase. The performance of each candidate landing point on the three major indicators is multiplied by the corresponding weight and then summed to obtain the comprehensive score of each candidate landing point, thereby forming complete candidate landing point score information.

[0106] Specifically, a reinforcement learning model based on the Actor-Critic structure can be used to systematically integrate the key evaluation dimensions covered by stability indicators, energy consumption indicators, and obstacle crossing indicators, so as to achieve a comprehensive quantitative score of candidate landing points and form score information. In this process, the obstacle-crossing metric focuses on terrain adaptability and accessibility, used for terrain adaptability evaluation. The model uses a convolutional feature extraction network or a point cloud feature encoder to calculate geometric features such as slope, height difference, and surface normal direction around the candidate landing point. These features are input into the Critic network to estimate the safety and supportability of the landing point. Areas with flat terrain, gentle slope, and high friction coefficient will receive higher obstacle-crossing related score weights. The energy consumption metric focuses on energy efficiency and torque smoothness, used for dynamic constraints and executability evaluation. The Actor network predicts whether the landing point is within the reachable range based on the robot's current body posture, joint kinematic range, leg length reachability, and other data. It also evaluates whether the torque and speed required to reach the target landing point from the current state are within the motor and structural capabilities, ensuring that the landing point is safely executable at the dynamic level and that energy consumption is reasonable. The relevant evaluation results are converted into the scoring criteria for the energy consumption metric. The stability metric focuses on ensuring posture stability, support stability, and anti-slip stability, used for global stability and gait consistency evaluation. When scoring, the network considers the robot's overall support polygon changes, center of gravity projection position, and expected gait phase to ensure that the foot placement distribution meets the walking rhythm and stability boundary conditions. Simultaneously, the balanced reward and gait phase consistency constraints introduced into the reward function further calibrate the scoring weights of the stability indicators, ensuring that foot placement selection balances local safety and global coordination. Finally, the Actor network generates action proposals for foot placement selection, while the Critic network provides stability and constraint feedback, dynamically adjusting and scientifically weighting the scoring criteria for the three indicators. The evaluation results of the three indicators are then weighted and integrated to form a comprehensive candidate foot placement score that reflects the overall performance of candidate foot placements in terms of terrain adaptability, dynamic executability, global stability, and energy efficiency.

[0107] For example, the formula for calculating the candidate foot placement score is as follows:

[0108] In the formula, This represents the final total reward value for reinforcement learning, used to comprehensively evaluate the merits of the landing point selection and guide the reinforcement learning model to optimize its strategy. This indicates a pre-set parameter used to adjust the overall stability reward weight, which can be adjusted in a course-based manner according to the training phase; This represents the overall stability reward value; This indicates that the parameters are preset to adjust the weight of the overall energy consumption reward, and can be adjusted in a course-based manner according to the training phase; This represents the comprehensive energy consumption reward value; This indicates that the parameters are pre-set to adjust the weight of the overall obstacle-crossing reward, and can be adjusted in a course-based manner according to the training phase; This represents the overall obstacle-crossing reward value.

[0109] Step S50: Based on the candidate landing point scoring information, determine the target landing point of the robot from the candidate landing points to complete the robot's landing point planning.

[0110] It should be noted that the target landing point is the landing position with the best comprehensive score selected from all candidate landing points. It not only meets the robot's stable movement requirements under the current terrain and effectively avoids instability, slippage and other situations, but also has high energy utilization efficiency, reducing unnecessary energy consumption and mechanical wear. At the same time, it can adapt well to obstacles or terrain with elevation differences, ensuring that the robot can successfully cross obstacles and continue to move.

[0111] It should be understood that when determining the target landing point, the scoring information of all candidate landing points is first systematically sorted, arranged in descending order of comprehensive score. During the sorting process, it is necessary to determine whether the score corresponding to the stability index in the scoring information meets the preset safety threshold, ensuring that the selected target landing point will not cause the robot to tip over, slip, or become unstable. Simultaneously, the scores of energy consumption and obstacle-crossing indicators are considered to avoid excessive energy consumption affecting battery life or inability to adapt to terrain leading to obstruction. The candidate landing point with the highest comprehensive score that meets all basic motion requirements is determined as the robot's target landing point. Once the target landing point is determined, the robot's landing point planning process is complete.

[0112] This embodiment provides a method for robot landing point planning. By comprehensively acquiring terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data, candidate landing points are generated based on this data. This ensures the physical accessibility of the landing points while avoiding invalid areas based on actual terrain conditions. Three key indicators—stability, energy consumption, and obstacle clearance—are constructed to comprehensively evaluate candidate landing points from different core dimensions. The scores calculated based on these three indicators objectively reflect the quality of the landing points. The final selected target landing point achieves the optimal combination of stability, energy efficiency, and terrain adaptability. This approach effectively improves the robot's ability to navigate complex terrain, reduces the risk of falls, and is flexibly applicable to various mobile platforms, providing an efficient and intelligent landing point planning solution for autonomous movement and operations in multiple scenarios.

[0113] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S20 may include steps S21 to S22: Step S21: Obtain robot gait parameters and robot posture prediction information based on the robot state data; It should be noted that robot gait parameters describe the robot's walking pattern, including stride length, stride frequency, gait phase, leg lift height, and support or swing time. These parameters determine the rhythm and range of the robot's leg movements. Robot posture prediction information, based on the robot's motion commands, current velocity, acceleration, and terrain-related information, infers the robot's posture at the moment the foot is about to land, thus defining the reachable domain of the legs.

[0114] It should be understood that by extracting walking-related data from the robot's state data and analyzing this data to determine the robot's current gait parameters, and by combining the robot's motion commands, current speed, acceleration and other information, the robot's posture when the foot lands can be comprehensively inferred. This provides information on robot gait parameters and robot posture prediction, and provides a basis for generating subsequent candidate foot landing points.

[0115] Step S22: Generate candidate landing points for the robot based on the terrain perception data, the robot gait parameters, and the robot posture prediction information.

[0116] It should be noted that terrain perception data is data reflecting the terrain conditions around the robot, including geometric features such as slope, height difference, surface normal direction, and support area, which can provide terrain environment reference for the generation of landing points.

[0117] It should be understood that the target swing leg is first determined based on the gait phase in the robot's gait parameters. Then, the nominal footing point in the ideal state under flat terrain, i.e. the desired footing point position, is calculated based on gait parameters such as stride length. Combining the robot's posture prediction information and the robot's current posture, as well as the joint kinematic range and leg length limit of the swing leg, the physically reachable search area is determined with the nominal footing point as the center. Finally, with reference to the terrain perception data, candidate footing points of the robot are generated within this search area through an appropriate sampling method.

[0118] In one feasible implementation, step S22 may include: The target swing leg and desired foot placement position of the robot are determined based on the robot's gait parameters. Based on the robot posture prediction information and the target swing leg, a physically reachable search area centered on the desired foot landing point is determined. Sampling is performed within the physically reachable search area to obtain multiple candidate landing points.

[0119] It should be noted that the target swing leg is the leg in the robot's current gait phase that needs to find a foothold. It is determined based on the gait phase in the gait parameters, and different gait phases correspond to different swing legs. The desired foothold position is a reference point in an ideal state on flat terrain, calculated based on the robot's current posture, velocity, and desired overall motion target. This position does not consider details such as small stones or potholes in the local terrain.

[0120] Additionally, the physically reachable search area is the physical reachable range of the target swing leg, which can be elliptical or rectangular in shape, and its range is determined by the robot's motion constraints. Robot posture prediction information is the inferred body posture of the robot when the foot is about to land, including the center of mass position, pitch angle, roll angle, etc., which affects the definition of the leg's reachable range.

[0121] In addition, sampling is the process of selecting potential landing points from the physically reachable search area. The methods used include grid sampling, random sampling, or sampling with specific rules that are equidistantly distributed along the robot's direction of travel. Different sampling methods can be adapted to different terrain scenarios.

[0122] It should be understood that extracting gait phase information from the robot's gait parameters allows us to determine which leg is currently in the swinging phase requiring a foothold, thus identifying the target swinging leg. Combining this with data such as the expected stride length from the gait parameters, we calculate the ideal foothold location, providing a reference benchmark for subsequent search area delineation. Subsequently, by combining robot posture prediction information and the current body posture, we clarify the joint kinematic range and leg length constraints of the target swinging leg. Using the previously determined desired foothold location as the center, these kinematic constraints are transformed into a specific spatial range, thus delineating the physically reachable search area that the target swinging leg can access, ensuring that subsequently sampled footholds are physically feasible. Based on the actual terrain complexity and planning efficiency requirements, an appropriate sampling method is selected, uniformly or selectively choosing multiple points within the delineated physically reachable search area. These points must cover the main feasible areas within the region, ultimately forming multiple candidate footholds to provide selection options for subsequent comprehensive scoring and screening.

[0123] This embodiment provides a method for robot foothold planning. It extracts robot gait parameters and robot posture prediction information from robot state data, ensuring that the core basis for generating candidate footholds is accurate and targeted, avoiding unfounded range delineation. By combining terrain perception data, robot gait parameters, and robot posture prediction information, it considers both the actual environmental conditions of the terrain and the robot's motion constraints and posture characteristics. The generated candidate footholds are both physically accessible and adapted to the terrain conditions, providing a high-quality candidate set for subsequent optimal foothold selection based on multiple indicators, further improving the rationality, safety, and feasibility of foothold planning.

[0124] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the landing point planning method of the robot in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0125] This application also provides a robot foot placement planning device, please refer to... Figure 3 The robot's landing point planning device includes: Data acquisition module 10 is used to acquire terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; Data processing module 20 is used to generate candidate landing points for the robot based on the terrain perception information and the robot state data; The data processing module 20 is also used to construct stability indicators, energy consumption indicators, and obstacle crossing indicators of the candidate landing points based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data. The landing point scoring module 30 is used to determine candidate landing point scoring information based on the stability index, the energy consumption index and the obstacle crossing index. The landing point determination module 40 is used to determine the target landing point of the robot from the candidate landing points based on the candidate landing point scoring information, so as to complete the landing point planning of the robot.

[0126] In one embodiment, the data processing module 20 is further configured to acquire the robot's base posture angle information, centroid projection information, foot state data, joint state data, and angular velocity based on the robot state data; Based on the base attitude angle information, the centroid projection information, and the foot state data, a stability index for the candidate foot landing point is constructed. The energy consumption index of the candidate landing point is constructed based on the joint state data and the angular velocity. The obstacle crossing index of the candidate landing point is constructed based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data.

[0127] In one embodiment, the foot state data includes foot support area information and foot tangential velocity; the data processing module 20 is further configured to determine attitude stability reward based on the base attitude angle information; The support margin bonus is determined based on the centroid projection information and the foot support area information. The anti-slip stability bonus is determined based on the foot tangential velocity. The stability index of the candidate foot placement point is constructed based on the posture stability reward, the support margin reward, and the anti-slip stability reward.

[0128] In one embodiment, the joint state data includes the number of joints, the current joint torque, the historical joint torque, and the torque acquisition time interval; the data processing module 20 is further configured to determine a power consumption reward based on the current joint torque, the angular velocity, and the number of joints. A torque smoothing reward is determined based on the current joint torque, the historical joint torque, and the torque acquisition time interval. The energy consumption index of the candidate landing point is constructed based on the power energy consumption reward and the torque smoothing reward.

[0129] In one embodiment, the terrain perception information includes obstacle height information, the height variance of the candidate landing area, the slope angle, and the normal direction; the motion accessibility constraint data includes the robot's leg swing height and the leg reachability range; the data processing module 20 is further configured to determine the leg swing space reward based on the obstacle height information and the leg swing height. The terrain support reward is determined based on the height variance, the slope angle, and the normal. No-collision bonuses are determined based on collision risk assessment data. The leg reachability reward is determined based on the coordinates of the candidate foot placement point and the leg reachability range. The obstacle-crossing index of the candidate landing point is constructed based on the leg swing space reward, the terrain support reward, the collision-free reward, and the leg reachability reward.

[0130] In one embodiment, the data processing module 20 is further configured to obtain robot gait parameters and robot posture prediction information based on the robot state data; Candidate landing points for the robot are generated based on the terrain perception data, the robot gait parameters, and the robot posture prediction information.

[0131] In one embodiment, the data processing module 20 is further configured to determine the target swing leg and desired foot placement position of the robot based on the robot gait parameters; Based on the robot posture prediction information and the target swing leg, a physically reachable search area centered on the desired foot landing point is determined. Sampling is performed within the physically reachable search area to obtain multiple candidate landing points.

[0132] The robot landing point planning device provided in this application, employing the robot landing point planning method described in the above embodiments, can solve the technical problem of poor stability and environmental adaptability of existing robots when traversing complex terrain. Compared with the prior art, the beneficial effects of the robot landing point planning device provided in this application are the same as those of the robot landing point planning method provided in the above embodiments, and other technical features in the robot landing point planning device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0133] This application provides a robot landing point planning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the robot landing point planning method in the above embodiment 1.

[0134] The following is for reference. Figure 4 The diagram illustrates a structural schematic of a robot foot placement planning device suitable for implementing embodiments of this application. The robot foot placement planning device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 4 The robot landing point planning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0135] like Figure 4As shown, the robot's landing point planning device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the robot's landing point planning device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the robot's landing point planning device to communicate wirelessly or wiredly with other devices to exchange data. Although a robot's landing point planning device with various systems is shown in the figure, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0136] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0137] The robot landing point planning device provided in this application, employing the robot landing point planning method described in the above embodiments, can solve the technical problem of poor stability and environmental adaptability of existing robots when traversing complex terrain. Compared with the prior art, the beneficial effects of the robot landing point planning device provided in this application are the same as those of the robot landing point planning method provided in the above embodiments, and other technical features of the robot landing point planning device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0138] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0140] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the robot's foot placement planning method in the above embodiments.

[0141] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory), or flash memory, optical fiber, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0142] The aforementioned computer-readable storage medium may be included in the robot's landing point planning device; or it may exist independently and not be installed in the robot's landing point planning device.

[0143] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the robot's landing point planning device, the robot's landing point planning device: acquires terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; generates candidate landing points for the robot based on the terrain perception information and robot state data; constructs stability indicators, energy consumption indicators, and obstacle crossing indicators for the candidate landing points based on the terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; determines candidate landing point scoring information based on the stability indicators, energy consumption indicators, and obstacle crossing indicators; and determines the robot's target landing point from the candidate landing points based on the candidate landing point scoring information, thereby completing the robot's landing point planning.

[0144] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0145] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0146] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0147] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the landing point planning method of the robot described above. This solves the technical problem of poor stability and environmental adaptability of existing robots when traversing complex terrain. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the landing point planning method of the robot provided in the above embodiments, and will not be repeated here.

[0148] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the robot foot placement planning method described above.

[0149] The computer program product provided in this application can solve the technical problem of poor stability and environmental adaptability of existing robots when traversing complex terrain. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the robot foothold planning method provided in the above embodiments, and will not be repeated here.

[0150] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for planning the landing points of a robot, characterized in that, The robot's landing point planning method includes: Acquire terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data; Candidate landing points for the robot are generated based on the terrain perception information and the robot state data; Based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data, the stability index, energy consumption index, and obstacle crossing index of the candidate landing point are constructed. The candidate landing point scoring information is determined based on the stability index, the energy consumption index, and the obstacle crossing index. Based on the candidate landing point scoring information, the target landing point of the robot is determined from the candidate landing points to complete the robot's landing point planning.

2. The method as described in claim 1, characterized in that, The steps of constructing stability, energy consumption, and obstacle-crossing indices for the candidate landing points based on the terrain perception information, robot state data, motion accessibility constraint data, and collision risk assessment data include: Based on the robot state data, obtain the robot's base posture angle information, center of mass projection information, foot state data, joint state data, and angular velocity; Based on the base attitude angle information, the centroid projection information, and the foot state data, a stability index for the candidate foot landing point is constructed. The energy consumption index of the candidate landing point is constructed based on the joint state data and the angular velocity. The obstacle crossing index of the candidate landing point is constructed based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data.

3. The method as described in claim 2, characterized in that, The foot state data includes information on the foot support area and the foot tangential velocity; The step of constructing the stability index of the candidate landing point based on the base attitude angle information, the centroid projection information, and the foot state data includes: The attitude stability reward is determined based on the base attitude angle information; The support margin bonus is determined based on the centroid projection information and the foot support area information. The anti-slip stability bonus is determined based on the foot tangential velocity. The stability index of the candidate foot placement point is constructed based on the posture stability reward, the support margin reward, and the anti-slip stability reward.

4. The method as described in claim 2, characterized in that, The joint status data includes the number of joints, current joint torque, historical joint torque, and torque acquisition time interval. The step of constructing the energy consumption index of the candidate landing point based on the joint state data and the angular velocity includes: The power consumption reward is determined based on the current joint torque, the angular velocity, and the number of joints. A torque smoothing reward is determined based on the current joint torque, the historical joint torque, and the torque acquisition time interval. The energy consumption index of the candidate landing point is constructed based on the power energy consumption reward and the torque smoothing reward.

5. The method as described in claim 2, characterized in that, The terrain perception information includes obstacle height information, height variance, slope angle, and normal of the candidate landing area; the motion accessibility constraint data includes the robot's leg swing height and leg reachability range. The step of constructing the obstacle-crossing index of the candidate landing point based on the terrain perception information, the motion accessibility constraint data, and the collision risk assessment data includes: The leg swing space reward is determined based on the obstacle height information and the leg swing height. The terrain support reward is determined based on the height variance, the slope angle, and the normal. No-collision bonuses are determined based on collision risk assessment data. The leg reachability reward is determined based on the coordinates of the candidate foot placement point and the leg reachability range. The obstacle-crossing index of the candidate landing point is constructed based on the leg swing space reward, the terrain support reward, the collision-free reward, and the leg reachability reward.

6. The method as described in claim 1, characterized in that, The step of generating candidate landing points for the robot based on the terrain perception information and the robot state data includes: Based on the robot state data, robot gait parameters and robot posture prediction information are obtained; Candidate landing points for the robot are generated based on the terrain perception data, the robot gait parameters, and the robot posture prediction information.

7. The method as described in claim 6, characterized in that, The step of generating candidate foot placement points for the robot based on the terrain perception data, the robot gait parameters, and the robot posture prediction information includes: The target swing leg and desired foot placement position of the robot are determined based on the robot's gait parameters. Based on the robot posture prediction information and the target swing leg, a physically reachable search area centered on the desired foot landing point is determined. Sampling is performed within the physically reachable search area to obtain multiple candidate landing points.

8. A robot foot placement planning device, characterized in that, The device includes: The data acquisition module is used to acquire terrain perception information, robot status data, motion accessibility constraint data, and collision risk assessment data. The data processing module is used to generate candidate landing points for the robot based on the terrain perception information and the robot state data; The data processing module is also used to construct stability indicators, energy consumption indicators, and obstacle crossing indicators for the candidate landing points based on the terrain perception information, the robot state data, the motion accessibility constraint data, and the collision risk assessment data. The landing point scoring module is used to determine candidate landing point scoring information based on the stability index, the energy consumption index, and the obstacle crossing index. The landing point determination module is used to determine the target landing point of the robot from the candidate landing points based on the candidate landing point scoring information, so as to complete the landing point planning of the robot.

9. A robot foot placement planning device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the robot's foot placement planning method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the robot's foot placement planning method as described in any one of claims 1 to 7.

Images