A method and apparatus for determining a ground contact state, a robot, and a storage medium

Abstract

The application provides a ground contact state determination method and device, a robot and a storage medium. The method comprises the following steps: determining the reaction force of the foot of the robot on the ground; estimating the probability of stable contact between the foot and the ground under the condition that the reaction force is known according to a pre-trained probability model; and obtaining the ground contact state of the foot and the ground according to the probability. The method can reduce the dependence of ground contact state detection on a contact sensor.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robots, and more particularly, to a method and device for determining a ground contact state, a robot, and a storage medium. BACKGROUND

[0002] At present, in order to detect the contact state of the foot of a robot with the ground, an additional contact sensor needs to be installed on the sole of the robot. However, due to the movement mode of the robot, the environment, and other objective factors, the contact sensor is usually prone to damage. Therefore, how to reduce the dependence of the detection of the ground contact state on the contact sensor has become a technical problem to be solved. SUMMARY

[0003] The present application provides a method and device for determining a ground contact state, a robot, and a storage medium, which can reduce the dependence of the detection of the ground contact state on a contact sensor.

[0004] In a first aspect, a method for determining a ground contact state is provided, which includes determining the reaction force of the foot of a robot on the ground, estimating the probability of stable contact of the foot with the ground in the case of known reaction force according to a pre-trained probability model, and obtaining the ground contact state of the foot with the ground according to the probability.

[0005] In the above technical solution, by determining the reaction force of the foot of a robot on the ground, the probability of stable contact of the foot of the robot with the ground in the case of known reaction force is estimated based on a pre-trained probability model, so as to directly obtain the ground contact state of the foot of the robot with the ground based on the probability, without relying on the detection of a contact sensor, thereby enabling the detection of the ground contact state of the foot of the robot with the ground without relying on the contact sensor as much as possible.

[0006] In combination with the first aspect, in some possible implementation manners, the probability model is trained in the following manner: a plurality of reaction forces of the foot of the robot on the ground and contact labels corresponding to the plurality of reaction forces are collected under different movement modes of the robot; the contact labels include a first label for representing stable contact of the foot with the ground and a second label for representing unstable contact of the foot with the ground; and the probability model is trained according to the plurality of reaction forces and the contact labels corresponding to the plurality of reaction forces.

[0007] In the technical solution, the multiple reaction forces of the robot's foot on the ground in different movement modes and the contact tags corresponding to the multiple reaction forces are collected, so that the probability model trained based on the data can accurately detect the ground contact state of the robot in different movement modes, that is, the probability model can be applied to the detection of the ground contact state of the robot in different movement modes, and the application range of the probability model is improved.

[0008] In combination with the first aspect, in some possible implementation manners, the training of the probability model according to the multiple reaction forces and the contact tags corresponding to the multiple reaction forces includes: training the probability model on a collected data set by maximum likelihood estimation; and the data set includes the multiple reaction forces and the contact tags corresponding to the multiple reaction forces.

[0009] In combination with the first aspect, in some possible implementation manners, the expression of the probability model is as follows:

[0010]

[0011] wherein p t (L=1|F t ) is the probability that the foot is in stable contact with the ground when F t is known, F t is the reaction force of the foot on the ground, represents the normal component of F t , and β and β0 are weight parameters.

[0012] In combination with the first aspect, in some possible implementation manners, the determination of the reaction force of the robot's foot on the ground includes: obtaining the joint position and joint speed of the robot; and determining the reaction force of the robot's foot on the ground according to the joint position and joint speed and the forward kinematics equation of the robot.

[0013] In combination with the first aspect, in some possible implementation manners, before the determination of the reaction force of the robot's foot on the ground, the method further includes: determining whether the robot's foot is provided with a contact sensor; in the case where the robot's foot is provided with the contact sensor, determining the ground contact state of the robot's foot on the ground by the contact sensor; and in the case where the robot's foot is not provided with the contact sensor, determining the reaction force of the robot's foot on the ground.

[0014] In the technical solution, the determination of whether the robot's foot is provided with a contact sensor is added, and different detection modes of the ground contact state are provided based on different determination results, so that the flexibility of the ground contact state detection is improved.

[0015] With reference to the first aspect, in some possible implementations, the robot is a multi-legged robot, the probability of the foot being in stable contact with the ground includes probabilities of multiple feet of the multi-legged robot being in stable contact with the ground respectively, and the method further includes: determining a speed of each leg of the multi-legged robot; and determining a speed of the multi-legged robot according to the speed of each leg of the multi-legged robot and the probabilities of the multiple feet of the multi-legged robot being in stable contact with the ground respectively.

[0016] The second aspect provides a device for ground contact state estimation, which includes: a determination module configured to determine a reaction force of a foot of a robot on a ground; an estimation module configured to estimate, according to a pre-trained probability model, a probability of the foot being in stable contact with the ground in a case where the reaction force is known; and an acquisition module configured to acquire a ground contact state of the foot on the ground according to the probability.

[0017] With reference to the second aspect, in some possible implementations, the device further includes: a training module configured to collect multiple reaction forces of a foot of a robot on a ground in different movement modes and contact labels corresponding to the multiple reaction forces respectively; wherein the contact labels include a first label used to represent that the foot is in stable contact with the ground and a second label used to represent that the foot is not in stable contact with the ground; and the probability model is trained according to the multiple reaction forces and the contact labels corresponding to the multiple reaction forces respectively.

[0018] With reference to the second aspect, in some possible implementations, the training module is specifically configured to train the probability model on a collected data set by maximum likelihood estimation; and the data set includes the multiple reaction forces and the contact labels corresponding to the multiple reaction forces respectively.

[0019] With reference to the second aspect, in some possible implementations, an expression of the probability model is as follows:

[0020]

[0021] wherein p t (L=1|F t ) is the probability of the foot being in stable contact with the ground in a case where F t is known, F t is the reaction force of the foot on the ground, represents a normal component of F t , and β and β0 are weight parameters.

[0022] In some possible implementation manners, in combination with the second aspect, the determining module is specifically configured to: acquire joint positions and joint velocities of the robot; and determine the reaction force of the foot of the robot on the ground according to the joint positions and the joint velocities and a forward kinematics equation of the robot.

[0023] In some possible implementation manners, in combination with the second aspect, the apparatus further includes a judging module configured to judge whether the foot of the robot is provided with a contact sensor; and the determining module is specifically configured to: in a case where the foot of the robot is provided with the contact sensor, determine the ground contact state of the foot of the robot on the ground by using the contact sensor; and in a case where the foot of the robot is not provided with the contact sensor, determine the reaction force of the foot of the robot on the ground.

[0024] In some possible implementation manners, in combination with the second aspect, the robot is a multi-legged robot, and the probability of the foot being in stable contact with the ground includes probabilities of a plurality of feet of the multi-legged robot being in stable contact with the ground respectively; and the apparatus further includes a body velocity determining module configured to determine a velocity of each leg of the multi-legged robot; and determine a velocity of the multi-legged robot according to the velocity of each leg of the multi-legged robot and the probabilities of the plurality of feet of the multi-legged robot being in stable contact with the ground respectively.

[0025] In a third aspect, a robot is provided, including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to invoke and run the executable program code from the memory, so that the robot performs the method in the first aspect or any possible implementation manner of the first aspect.

[0026] In a fourth aspect, a computer program product is provided, which includes computer program code. When the computer program code is run on a computer, the computer program code causes the computer to perform the method in the first aspect or any possible implementation manner of the first aspect.

[0027] In a fifth aspect, a computer-readable storage medium is provided, which stores computer program code. When the computer program code is run on a computer, the computer program code causes the computer to perform the method in the first aspect or any possible implementation manner of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a schematic flowchart of a method for determining a ground contact state according to an embodiment of the present application;

[0029] Figure 2 is a schematic flowchart of another method for determining a ground contact state according to an embodiment of the present application; is a schematic flowchart of a method for determining a ground contact state according to an embodiment of the present application;

[0030] Figure 3 is a structural schematic diagram of a touch-down state determination device provided by an embodiment of the present application;

[0031] Figure 4 is a structural schematic diagram of a robot provided by an embodiment of the present application. DETAILED DESCRIPTION

[0032] The technical solutions in the present application will be described in detail below with reference to the drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " represents the meaning of or, for example, A / B can represent A or B: "and / or" in the text only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, A and B together, and B alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0033] Hereinafter, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more features.

[0034] Robots are usually expected to be able to perform real-time operations in places that are difficult for humans to reach, such as uninhabited areas, mountainous areas, underground environments, etc. This requires accurate real-time estimation of information such as position, speed, direction, etc. of the robot, and the robot often encounters situations such as foot slipping and uneven ground. Therefore, the traditional assumption that the robot foot, i.e. the foot, is always in stable contact with the ground without slipping is often not valid.

[0035] Almost all robot state estimation, motion planning, etc. rely on the function implementation of the real-time state (such as position, speed, direction, etc.) of the robot, which will use the leg kinematics of the robot and the Leg Odometry (LO) technology. The LO technology can extract the body speed from the leg speed (relative to the robot body coordinate system) of the robot. The input of the LO technology usually includes joint encoders and leg kinematics models. However, the LO technology relies on at least one foot of the robot being in stable contact with the ground. However, as mentioned above, the assumption that the robot foot is always in stable contact with the ground without slipping is often not valid.

[0036] At present, in order to detect the contact state of the foot of the robot with the ground, that is, the ground contact state of the foot of the robot, an additional contact sensor needs to be installed on the sole of the robot, but due to the movement mode of the robot, the environment and other objective factors, the contact sensor is usually easy to be damaged. Therefore, how to reduce the dependence of the ground contact state detection on the contact sensor has become a technical problem to be solved.

[0037] To solve the above technical problems, the embodiment of the present application provides a ground contact state determination method, which is beneficial to reduce the dependence of the ground contact state detection on the contact sensor, so as to reduce the hardware cost to a certain extent. The above ground contact state determination method can be applied to a robot, which can be a multi-legged robot, and the robot can be integrated with a control module, which can include a ground contact state detection algorithm. The detection algorithm can be implemented through the flow shown in the following Figure 1 Through the method in the embodiment, the ground contact state of the foot of the robot with the ground can be detected as much as possible without relying on the contact sensor.

[0038] Figure 1 is a schematic flowchart of a ground contact state determination method provided by the embodiment of the present application.

[0039] As shown in Figure 1 , the method comprises:

[0040] Step 101, determining the reaction force of the foot of the robot on the ground.

[0041] Step 102, according to a probability model obtained by pre-training, estimating the probability of stable contact between the foot and the ground under the condition of knowing the reaction force.

[0042] Step 103, according to the probability, obtaining the ground contact state of the foot and the ground.

[0043] In the embodiment shown in Figure 1 , by determining the reaction force of the foot of the robot on the ground, based on the probability model obtained by pre-training, the probability of stable contact between the foot of the robot and the ground under the condition of knowing the reaction force is estimated, so that the ground contact state of the foot of the robot and the ground is directly obtained based on the probability, without relying on the detection of the contact sensor, so that the ground contact state of the foot of the robot and the ground can be detected as much as possible without relying on the contact sensor.

[0044] The specific implementation mode of each step in the embodiment shown in Figure 1 will be described below:

[0045] In step 101, the reaction force of the foot of the robot on the ground is determined.

[0046] wherein, the ground reaction force (GRF) of the foot of the robot is the reaction force of the foot of the robot to the ground when the foot of the robot contacts the ground.

[0047] It can be understood that the stable stance depends on the friction force applied by the foot of the robot to the ground, which is proportional to the normal component of the ground reaction force GRF and the static friction coefficient between the foot and the ground. Generally, the static friction coefficient between the foot and the ground is unknown, and it is difficult to directly estimate the GRF when the foot is not equipped with a contact sensor. Therefore, in the embodiment, a method for determining the GRF is provided as follows when the foot of the robot is not equipped with a contact sensor:

[0048] First, the joint position and joint velocity of the robot are obtained, and then the ground reaction force of the foot of the robot is determined according to the joint position and joint velocity and the forward kinematics equation of the robot. In the embodiment, the joint position and joint velocity of the robot can be detected based on the position sensor and velocity sensor arranged at the joints of the robot.

[0049] For example, the GRF can be calculated by the following formula:

[0050]

[0051] wherein, p is the joint position, is the joint velocity, J(p) is the partial derivative of the forward kinematics equation of the robot, g b represents the acceleration of gravity, is a vector representing the centrifugal force, Coriolis force, and gravity, and τ is the control torque of the robot, which can be understood as the control torque of the joint of the robot. The forward kinematics (FK) equation of the robot leg can be summarized as follows:

[0052] p f = FK(φ, ρ)

[0053] wherein, p f is the position of the foot in the robot body coordinate system, φ is the joint angle of all joints of the robot leg, and ρ represents the generalized kinematics parameters, including the length of the connecting rod between the joints, the shaft offset, etc.

[0054] In step 102, the probability of the foot being in stable contact with the ground is estimated according to the pre-trained probability model, in the case that the reaction force is known. Thus, in step 103, the ground contact state of the foot with the ground can be obtained based on the probability of the foot being in stable contact with the ground. The ground contact state of the foot with the ground can be a state of the foot being in stable contact with the ground or a state of the foot being in unstable contact with the ground. If the estimated probability of the foot being in stable contact with the ground is greater than or equal to a preset probability threshold, it can be determined that the ground contact state is the state of the foot being in stable contact with the ground. If the estimated probability of the foot being in stable contact with the ground is less than the preset probability threshold, it can be determined that the ground contact state is the state of the foot being in unstable contact with the ground.

[0055] For example, the preset probability threshold can be set according to actual needs. For example, the higher the detection accuracy requirement of the ground contact state, the greater the probability threshold can be set. The lower the detection accuracy requirement of the ground contact state, the smaller the probability threshold can be set. Optionally, the probability threshold can be greater than or equal to 0.5, for example, 0.6 or 0.7. However, the present embodiment is not limited in this regard.

[0056] For example, the preset probability threshold can also be determined according to the motion state of the robot in a period of time. If the motion state of the robot is stable motion, for example, walking on a flat ground, the probability threshold can be relatively small. If the motion state of the robot is unstable motion, for example, running or jumping on the ground, the probability threshold can be relatively large. That is, the robot can pre-store a first probability threshold and a second probability threshold, the first probability threshold being greater than the second probability threshold. If the current motion state of the robot is stable motion, the ground contact state of the foot with the ground can be obtained according to the second probability threshold and the estimated probability of the foot being in stable contact with the ground. If the estimated probability of the foot being in stable contact with the ground is greater than or equal to the second probability threshold, it can be determined that the ground contact state of the foot with the ground is the state of stable contact. Otherwise, it can be determined that the ground contact state of the foot with the ground is the state of unstable contact. If the current motion state of the robot is unstable motion, the ground contact state of the foot with the ground can be obtained according to the first probability threshold and the estimated probability of the foot being in stable contact with the ground. If the estimated probability of the foot being in stable contact with the ground is greater than or equal to the first probability threshold, it can be determined that the ground contact state of the foot with the ground is the state of stable contact. Otherwise, it can be determined that the ground contact state of the foot with the ground is the state of unstable contact.

[0057] By setting different probability thresholds for different motion states, and setting a smaller probability threshold for stable motion and a larger probability threshold for unstable motion, it is beneficial to accurately estimate the ground contact state.

[0058] In this embodiment, the ground reaction force GRF of the foot of the robot can be taken as an input of the probability model, so that the probability model can output the probability of stable contact between the foot and the ground under the known ground reaction force. The probability of stable contact between the foot and the ground can reflect the actual contact state of the foot and the ground.

[0059] In an exemplary embodiment, the probability model can be trained by S1-S2 as follows:

[0060] S1, collect a plurality of ground reaction forces of the foot of the robot under different movement modes and a plurality of contact labels corresponding to the plurality of ground reaction forces, respectively.

[0061] S2, train the probability model according to the plurality of ground reaction forces and the plurality of contact labels corresponding to the plurality of ground reaction forces, respectively.

[0062] In this embodiment, by collecting a plurality of ground reaction forces of the foot of the robot under different movement modes and a plurality of contact labels corresponding to the plurality of ground reaction forces, respectively, the probability model trained based on these data can accurately detect the contact state of the robot under different movement modes, i.e., the probability model can be applied to the detection of the contact state of the robot under different movement modes.

[0063] In S1, the contact label includes a first label for indicating stable contact between the foot and the ground and a second label for indicating unstable contact between the foot and the ground. For example, the first label can be represented by L=1, and the second label can be represented by L=0. S1 can be understood as collecting a plurality of data samples for training the probability model and labeling the data samples with sample labels. The plurality of data samples can be understood as the plurality of ground reaction forces of the foot under different movement modes, and the sample labels can be understood as the contact labels.

[0064] In a specific implementation, in order to further make the applicability of the trained probability model more extensive, when collecting data, a plurality of ground reaction forces of the foot of the robot under different movement modes on different types of ground can be collected. The different types of ground can include grass, mountain road, highway, etc.

[0065] For example, when determining the contact labels corresponding to the plurality of ground reaction forces, respectively, a contact sensor can be used. The contact sensor can output the contact labels corresponding to the plurality of ground reaction forces, respectively, to indicate the contact state of the foot of the robot represented by different ground reaction forces of the foot of the robot.

[0066] In the S2, the plurality of reaction forces collected in the S1 and the contact labels corresponding to the plurality of reaction forces respectively can be taken as a data set to train the probability model on the data set.

[0067] For example, the S2 can be implemented by training the probability model on the collected data set by maximum likelihood estimation, the data set including the plurality of reaction forces collected and the contact labels corresponding to the plurality of reaction forces respectively.

[0068] For example, the probability model can be expressed as follows:

[0069]

[0070] wherein p t (L=1|F t ) is the probability of the foot being in stable contact with the ground given F t , F t is the reaction force of the foot on the ground, represents the normal component of F t , and β and β0 are weight parameters. The probability model can be understood as a probability model of the foot of the robot being in stable contact with the ground.

[0071] In a specific implementation, the plurality of reaction forces of the foot of the robot on the ground in different movement modes and the contact labels corresponding to the plurality of reaction forces respectively can be brought into the expression of the probability model to train the β and β0. When the β and β0 are trained, the unknown quantity in the expression of the probability model is the normal component of F t , i.e. Therefore, after the current reaction force of the foot of the robot on the ground at the current time is obtained, the normal component of the current reaction force can be determined, and then the normal component of the current reaction force is brought into the expression of the probability model to calculate the probability of the foot being in stable contact with the ground given the current reaction force, and then determine the contact state of the foot of the robot on the ground at the current time based on the probability.

[0072] In the embodiment, the theoretical basis for determining whether the foot of the robot is in stable contact with the ground is that a stable foothold depends on the friction force exerted by the foot of the robot on the ground, and the friction force is proportional to the normal component of the ground reaction force GRF and the static friction coefficient of the foot-ground. Combined with the theoretical basis, the expression of the probability model can be constructed.

[0073] In the example embodiment, before determining the reaction force of the foot of the robot on the ground, the method further comprises: determining whether the foot of the robot is provided with a contact sensor; in the case that the foot of the robot is provided with the contact sensor, determining the ground contact state of the foot of the robot on the ground by the contact sensor; and in the case that the foot of the robot is not provided with the contact sensor, determining the reaction force of the foot of the robot on the ground.

[0074] Specifically, before performing the step 101, it can be determined whether the foot of the robot is provided with a contact sensor. If the foot of the robot is provided with the contact sensor, the contact label output by the contact sensor can be directly obtained. When the contact label output by the contact sensor is L=1, it can be determined that the ground contact state of the foot of the robot on the ground is a stable contact state. When the contact label output by the contact sensor is L=0, it can be determined that the ground contact state of the foot of the robot on the ground is an unstable contact state. That is, if it is determined that the foot of the robot is provided with the contact sensor, the detection of the ground contact state can be directly implemented based on the contact label output by the contact sensor without performing the steps 101 to 103. If it is determined that the foot of the robot is provided with the contact sensor, the steps 101 to 103 can be further performed to estimate the probability of the stable contact of the foot of the robot on the ground in the case of the known reaction force, so as to obtain the ground contact state of the foot on the ground in combination with the probability.

[0075] The embodiment can provide different detection manners of the ground contact state based on different determination results by adding the determination step of whether the foot of the robot is provided with the contact sensor, which is beneficial to improve the flexibility of the detection of the ground contact state.

[0076] In the example embodiment, the robot is a multi-legged robot, such as a biped robot, a quadruped robot, etc. The probability of the stable contact of the foot on the ground includes the probabilities of the stable contact of the multiple feet of the multi-legged robot on the ground respectively, and the method further comprises: determining the speed of each leg of the multi-legged robot; and determining the speed of the multi-legged robot according to the speed of each leg of the multi-legged robot and the probabilities of the stable contact of the multiple feet of the multi-legged robot on the ground respectively.

[0077] Specifically, the probability of the stable contact of each leg of the multi-legged robot on the ground can be determined by the steps 101 to 102. The speed of each leg of the multi-legged robot can be detected by the speed sensor arranged on each leg of the multi-legged robot. Thus, the speed of the multi-legged robot can be determined according to the speed of each leg of the multi-legged robot and the probabilities of the stable contact of the multiple feet of the multi-legged robot on the ground respectively, and the speed of the multi-legged robot can be understood as the speed of the body of the multi-legged robot.

[0078] For example, the speed of the multi-legged robot can be calculated by weighted average calculation of the speed of each leg of the multi-legged robot and the probability of stable contact of each foot part of the multi-legged robot with the ground.

[0079]

[0080] wherein, is the probability of stable contact of the foot of the ith leg of the multi-legged robot with the ground at the current time t, given the reaction force of the foot of the ith leg of the multi-legged robot with the ground is the speed of the ith leg of the multi-legged robot at the current time t, and i represents the index number of the leg of the multi-legged robot, and C represents the index number set of the legs of the multi-legged robot.

[0081] It should be noted that in the embodiment, the estimated probability of stable contact of the foot with the ground is used for estimating the body speed of the robot, and in specific implementation, it can also be applied to other tasks, such as real-time online state estimation, motion planning, etc.

[0082] In the exemplary embodiment, the flowchart of the determination method of the ground contact state can be referred to Figure 2 , which comprises:

[0083] Step 201: Start of ground contact state modeling. The start of ground contact state modeling can be understood as starting to detect the ground contact state. For example, a modeling start instruction can be sent to the robot to trigger the robot to start detecting the ground contact state.

[0084] Step 202: Determine whether the foot of the robot is provided with a contact sensor. If yes, execute step 203, otherwise execute step 204.

[0085] Step 203: Determine the ground contact state of the foot of the robot through the contact sensor. For example, the contact sensor can directly output a contact label, and if the output contact label is 1, it indicates that the ground contact state is a stable contact state of the foot with the ground, and if the output contact label is 0, it indicates that the ground contact state is an unstable contact state of the foot with the ground. That is, if the foot of the robot is provided with a contact sensor, the contact label is directly read from the contact sensor, and the contact label L=1 represents a stable contact or stable foothold, and the contact label L=0 represents an unstable contact or unstable foothold.

[0086] Step 204: Execute the algorithm to determine the contact label process. The algorithm to determine the contact label process can be steps 101 to 103 described above.​

[0087] Step 205: Use the obtained contact tags for subsequent tasks. These subsequent tasks could include, for example, estimating the robot's velocity, real-time online state estimation, and motion planning.

[0088] The ground contact detection method provided in this embodiment can detect both stable and unstable ground contact states regardless of whether contact sensors are installed on the robot's feet. This not only reduces hardware costs but also allows for easy integration as a module into various other multi-legged robot state estimation and motion planning frameworks / algorithms without requiring additional hardware, demonstrating strong portability and wide applicability.

[0089] The accompanying drawings are merely illustrative of the steps included in the method according to exemplary embodiments of this specification and are not intended to be limiting. It is readily understood that the steps shown in the drawings do not indicate or limit the temporal order of these steps. Furthermore, it is readily understood that these steps may be performed synchronously or asynchronously, for example, in multiple modules.

[0090] Figure 3 This is a schematic diagram of the structure of a device for determining the ground contact state provided in an embodiment of this application.

[0091] For example, such as Figure 3 As shown, the device 300 includes: a determining module 301 for determining the reaction force of the robot's foot on the ground; a predicting module 302 for predicting the probability of the foot making stable contact with the ground given the known reaction force, based on a pre-trained probability model; and an acquiring module 303 for acquiring the contact state of the foot with the ground based on the probability.

[0092] In one possible implementation, the device 300 further includes: a training module for collecting multiple reaction forces of the robot's feet on the ground under different movement modes and contact tags corresponding to the multiple reaction forces; wherein the contact tags include a first tag for characterizing stable contact between the feet and the ground and a second tag for characterizing unstable contact between the feet and the ground; and the probability model is trained based on the multiple reaction forces and the contact tags corresponding to the multiple reaction forces.

[0093] In one possible implementation, the training module is specifically used to train the above-mentioned probability model on the collected dataset through maximum likelihood estimation; wherein, the above-mentioned dataset includes the above-mentioned multiple reaction forces and the contact labels corresponding to the above-mentioned multiple reaction forces respectively.

[0094] In one possible implementation, the expression for the above probability model is as follows:

[0095]

[0096] Where, p t (L=1|F t ) is given F t In the above-mentioned case, the probability F of stable contact between the foot and the ground is... t The reaction force of the feet on the ground. Representing F t The normal component is , and β and β0 are weighting parameters.

[0097] In one possible implementation, the determining module 301 is specifically used to: obtain the joint position and joint velocity of the robot; and determine the reaction force of the robot's feet on the ground based on the joint position and joint velocity and the robot's forward kinematic equation.

[0098] In one possible implementation, the device 300 further includes: a judgment module for judging whether the robot's feet are equipped with contact sensors; and a determination module 301 specifically used to determine the contact state between the robot's feet and the ground by means of the contact sensors when the robot's feet are equipped with contact sensors; and to determine the reaction force of the robot's feet on the ground when the robot's feet are not equipped with contact sensors.

[0099] In one possible implementation, the robot is a multi-legged robot, and the probability of the legs making stable contact with the ground includes the probability of multiple legs of the multi-legged robot making stable contact with the ground respectively. The device 300 further includes: a body speed determination module, used to determine the speed of each leg of the multi-legged robot; and to determine the speed of the multi-legged robot based on the speed of each leg of the multi-legged robot and the probability of multiple legs of the multi-legged robot making stable contact with the ground respectively.

[0100] Figure 4 This is a schematic diagram of the structure of a robot provided in an embodiment of this application.

[0101] For example, such as Figure 4 As shown, the robot includes a memory 401 and a processor 402. The memory 401 stores executable program code, and the processor 402 is used to call and execute the executable program code to perform a method for determining the ground contact state.

[0102] The embodiment can divide the robot into functional modules according to the method examples described above. For example, each functional module can be divided, or two or more functions can be integrated into one processing module. The integrated module can be implemented in the form of hardware. It should be noted that the division of the modules in the embodiment is illustrative, and is only a logical functional division. In actual implementation, another division manner can be used.

[0103] In the case of dividing each functional module according to each function, the robot can include a determination module, an estimation module, an acquisition module, and the like. It should be noted that all related contents of each step involved in the method embodiment can be cited in the functional description of the corresponding functional module, and will not be described here.

[0104] The robot provided by the embodiment is used to execute the determination method of the touch-down state, and thus the same effect as the implementation method can be achieved.

[0105] In the case of using an integrated unit, the robot can include a processing module and a storage module. The processing module can be used to control and manage the actions of the robot. The storage module can be used to support the robot to execute program codes and data.

[0106] The processing module can be a processor or a controller, which can implement or execute various exemplary logical blocks, modules and circuits represented in combination with the disclosure of the present application. The processor can also be a combination of computing functions, such as one or more microprocessor combinations, a combination of digital signal processing (DSP) and microprocessor, and the like. The storage module can be a memory.

[0107] The embodiment further provides a computer readable storage medium, which stores computer program codes. When the computer program codes are run on a computer, the computer is caused to execute the related method steps to implement the determination method of the touch-down state in the embodiment.

[0108] The embodiment further provides a computer program product, which causes a computer to execute the related steps to implement the determination method of the touch-down state in the embodiment when the computer program product is run on the computer.

[0109] In addition, the robot provided by the embodiment of the present application can be a chip, an assembly or a module. The robot can include a connected processor and a memory. The memory is used to store instructions. When the robot is running, the processor can invoke and execute the instructions to enable the chip to execute the determination method of the touch-down state in the embodiment.

[0110] Among them, the robot, the computer readable storage medium, the computer program product or the chip provided by the embodiment are used for executing the corresponding method provided above, therefore, the beneficial effects that can be achieved are referable to the beneficial effects in the corresponding method provided above, and will not be repeated here.

[0111] Through the description of the above embodiments, those skilled in the art can understand that, for the convenience and brevity of description, only the division of the above functional modules is taken as an example, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0112] In the embodiments provided in the present application, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules or units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0113] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of determining a ground contact status, characterized by, The method comprises: determining the reaction force of the foot of the robot on the ground; estimating the probability of stable contact of the foot with the ground in the case of known reaction force according to a pre-trained probability model; obtaining the ground contact state of the foot with the ground according to the probability; wherein the probability model is trained in the following way: collecting multiple reaction forces of the foot of the robot on the ground in different movement modes and the contact labels corresponding to the multiple reaction forces respectively; wherein the contact labels include a first label for representing stable contact of the foot with the ground and a second label for representing unstable contact of the foot with the ground; training the probability model according to the multiple reaction forces and the contact labels corresponding to the multiple reaction forces respectively; the expression of the probability model is as follows: wherein, is the probability that the foot is in stable contact with the ground, is the reaction force of the foot on the ground, represents the normal component of the force, and is a weight parameter.​ 2. The method of claim 1, wherein, training the probability model according to the multiple reaction forces and the contact labels corresponding to the multiple reaction forces respectively comprises: training the probability model on the collected data set by maximum likelihood estimation; wherein the data set comprises the multiple reaction forces and the contact labels corresponding to the multiple reaction forces respectively.

3. The method of claim 1, wherein, The determination of the reaction force of the foot of the robot on the ground comprises: obtaining the joint position and joint velocity of the robot; determining the reaction force of the foot of the robot on the ground according to the joint position and joint velocity and the forward kinematics equation of the robot; The reaction force of the robot's foot on the ground This is calculated by the equation: wherein is the joint position, is the joint velocity, is the partial derivative of the forward kinematics equation of the robot, denotes the gravitational acceleration, is a vector determined based on centrifugal force, Coriolis force, gravitational force, is the control torque of the robot.

4. The method of claim 1, wherein, Before determining the reaction force of the foot of the robot on the ground, the method further comprises: judging whether the foot of the robot is provided with a contact sensor; in the case that the foot of the robot is provided with a contact sensor, determining the ground contact state of the foot of the robot with the ground through the contact sensor; in the case that the foot of the robot is not provided with a contact sensor, determining the reaction force of the foot of the robot on the ground.

5. The method according to any one of claims 1 to 4, characterized in that, The robot is a multi-legged robot, and the probability of stable contact of the foot with the ground comprises the probability of stable contact of each foot of the multi-legged robot with the ground, and the method further comprises: determining the speed of each leg of the multi-legged robot; determining the speed of the multi-legged robot according to the speed of each leg of the multi-legged robot and the probability of stable contact of each foot of the multi-legged robot with the ground.

6. An apparatus for touch-down state estimation, the apparatus comprising: The device comprises: a determination module for determining the reaction force of the foot of the robot on the ground; an estimation module for estimating the probability of stable contact of the foot with the ground in the case of known reaction force according to a pre-trained probability model; an obtaining module for obtaining the ground contact state of the foot with the ground according to the probability. A training module is configured to collect a plurality of reaction forces of a foot of a robot on a ground in different movement modes and corresponding contact tags of the plurality of reaction forces, wherein the contact tags include a first tag for representing stable contact of the foot on the ground and a second tag for representing unstable contact of the foot on the ground; and train the probabilistic model according to the plurality of reaction forces and the corresponding contact tags of the plurality of reaction forces. An expression of the probabilistic model is as follows: wherein, is the probability that the foot is in stable contact with the ground, is the reaction force of the foot on the ground, represents the normal component of the force, and is a weight parameter.​​ 7. A robot, characterized in that The robot comprises: a memory configured to store executable program codes; a processor configured to call and run the executable program codes from the memory, so that the robot performs the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, when the computer program is executed, the method according to any one of claims 1 to 5 is realized. The computer readable storage medium stores a computer program, when the computer program is executed, the method according to any one of claims 1 to 5 is realized.

Images