Six-dimensional force-controllable perception bionic foot system and control method

Abstract

The application discloses a six-dimensional force control perception bionic foot system and a control method, and relates to the technical field of bionic foot control. The system comprises a bionic foot main body, a parallel driving mechanism, an ankle joint platform, a multi-point six-dimensional force sensing device and an ankle joint controller. The multi-point six-dimensional force sensing device is used for acquiring information of the bionic foot main body and the ankle joint platform. The ankle joint controller is used for adjusting the parallel driving mechanism. The parallel driving mechanism controls the bionic foot main body via the ankle joint platform. The ankle joint controller fuses information transmitted by each six-dimensional force sensor and then controls the parallel driving mechanism to adjust the ankle joint platform and the bionic foot main body. The application combines the triangular arch structure of the foot arch shape with flexible bionic materials, can realize larger area contact and adhesion on uneven ground, fuses information of three six-dimensional force sensors through a model, calculates zero moment point data and the like, provides more sufficient feedback information for attitude control, introduces feedback information, forms a closed loop dynamic feedback and improves stability again.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of humanoid robots, and particularly relates to a six-dimensional force control perception bionic foot system and a control method. BACKGROUND

[0002] With the development of humanoid robots in complex terrain walking, environmental adaptability control and other aspects, the requirement for foot perception ability is increasingly improved. As the execution end of contact with the ground, the performance of foot structure plays a decisive role in gait control of the whole machine in stability, perception ability, adaptability and other aspects. The control of the ankle joint of the existing humanoid robot mostly depends on single degree of freedom servo driving or traditional series structure. The series driving mechanism, that is, using two motors to control the pitch and roll degrees of freedom of the ankle joint, the motors of the series mechanism are usually stacked, resulting in large height of the ankle joint, offset of the gravity center, unreasonable mass distribution, bulky volume, which is not conducive to the posture control of the robot.

[0003] In the prior art, the foot structure of the commonly used humanoid robot usually adopts a rigid flat bottom structure, which contacts with the ground through an integrated rigid sole. This kind of structure has the advantages of simple manufacturing, high structural rigidity and high force transmission efficiency, but lacks flexibility and adaptability. The traditional sole force sensing system mostly uses single point or two-dimensional / three-dimensional force sensors, which is difficult to accurately estimate the ground reaction force, contact posture and friction force at the same time, and cannot perceive the force distribution and landing state of multiple areas of the foot, resulting in limited perception accuracy of the ground reaction, control feedback lag, insufficient stability and limited perception ability. Limited by the structure and sensing layout, the foot posture adjustment usually depends on the upper control or visual information, and cannot realize rapid and real-time landing feedback and adjustment, resulting in poor stability in dynamic gait or impact.

[0004] In view of the above problems, some existing researches try to improve the foot perception ability by arranging multi-point sensors on the sole and designing flexible materials, but most of them are limited to two-dimensional pressure distribution sensors, and a closed-loop force control system integrating structure and control has not yet been formed. Some researches also propose parallel mechanisms with driven ankle joints, but lack efficient control methods integrated with sensing. In the combination of existing robot feet and sensors, the following problems often exist: 1) poor structural adaptability: the existing rigid flat bottom structure is difficult to adapt to uneven ground and has small ground contact area; 2) insufficient contact perception ability: traditional foot perception mainly relies on single point three-dimensional or six-dimensional force sensors, which is difficult to obtain real-time multi-area force and contact posture, and cannot accurately judge the landing state and friction boundary; 3) slow ankle joint response and non-closed-loop control: the series driving structure is large in size, slow in dynamic response, and lacks posture feedback adjustment mechanism of the real force state of the foot, resulting in poor landing impact buffering performance and poor gait stability, and the parallel driving lacks efficient control method integrated with multi-point sensing. SUMMARY

[0005] To solve the above three problems, the application aims to provide a six-dimensional force control perception bionic foot system and a control method. By adopting a triangular arch support structure with an arch shape and combining it with a flexible bionic foot bottom material, compared with a traditional rigid flat foot structure, a larger area of contact and adhesion on uneven ground can be achieved. Three six-dimensional force sensors are arranged to collect the spatial force information of the forefoot, the hind foot, and the ankle joint platform, respectively, and the zero moment point position, the ground normal inclination, and other data are calculated through a three-point moment model fusion, thereby providing more sufficient feedback information for attitude control. A symmetric link drive ankle joint is adopted to replace the traditional serial motor structure, which not only has a more compact structure, but also has a faster response speed, a smaller driving coupling, and enhanced ankle joint dynamic response and attitude adjustment capability. In addition, the gait stability and terrain adaptability are improved. Through the introduction of a three-layer architecture in the control method, combined with real-time sensing of the force information and gait recognition, the control strategy can be automatically adjusted according to the support phase or the swing phase, thereby realizing the whole process of closed-loop dynamic adjustment from impact buffering, stable support to starting and leaving the ground.

[0006] The technical scheme is implemented by:

[0007] A six-dimensional force control perception bionic foot system, comprising a bionic foot main body, a parallel driving mechanism, an ankle joint platform, a multi-point six-dimensional force sensing device, and an ankle joint controller. The multi-point six-dimensional force sensing device is used to obtain information of the bionic foot main body and the ankle joint platform. The ankle joint controller is used to adjust the parallel driving mechanism, and the parallel driving mechanism controls the bionic foot main body through the ankle joint platform. The bionic foot main body comprises a foot triangular arch structure and a foot bottom. The foot bottom is used to contact the ground and support the foot triangular arch structure. The ankle joint platform comprises a connected ankle joint and an ankle joint assembly. The ankle joint assembly is connected to the foot triangular arch structure. The multi-point six-dimensional force sensing device comprises an ankle joint assembly six-dimensional force sensor and two six-dimensional force sensors arranged on the foot bottom. The ankle joint controller adopts a multi-layer control architecture, including a fusion layer, a bottom layer, a middle layer, and a high layer. The ankle joint controller fuses the information transmitted by each six-dimensional force sensor and controls the parallel driving mechanism to adjust the ankle joint platform and the bionic foot main body based on the fusion result.

[0008] Preferably, the foot bottom comprises a forefoot, a hind foot, and a fascia connector. The fascia connector is used to connect the forefoot and the hind foot. The foot triangular arch structure adopts a rigid arch assembly to fix the bionic foot main body and the ankle joint platform. The rigid arch assembly comprises a front arch segment and a rear arch segment. The front arch segment is used to connect the forefoot and the ankle joint assembly. The rear arch segment is used to connect the hind foot and the ankle joint assembly.

[0009] Preferably, the three six-dimensional force sensors are arranged on the forefoot, the hind foot, and the ankle joint assembly, respectively, to transmit the real-time acquired arbitrary force and torque information to the ankle joint controller.

[0010] Preferably, the parallel driving mechanism comprises symmetrical connecting rods, two motors, a spherical hinge, a shank rod, a push rod, a spherical joint; the ankle joint assembly is connected to the symmetrical connecting rods through the spherical hinge; the symmetrical connecting rods drive the push rod based on the two motors, and the push rod and the symmetrical connecting rods are connected through the spherical joint; the two motors are installed inside the shank rod, and the shank rod connects the ankle joint assembly through the ankle joint.

[0011] Preferably, the spherical joint lower part and the symmetrical connecting rod are both provided with threads, and the spherical joint and the symmetrical connecting rod are connected based on the threads.

[0012] Preferably, the fusion layer fuses each force and torque information transmitted by each six-dimensional force sensor, and is used to estimate a plurality of parameters corresponding to the motion of the bionic foot body; the bottom layer is used to control each motor in the parallel driving mechanism; the middle layer adopts a posture correction controller; and the high layer is used to calculate a control strategy of the bionic foot and transmit the control strategy to the middle layer or the bottom layer.

[0013] Preferably, the plurality of parameters estimated by the fusion layer include total support force, total torque, contact posture and zero torque point distribution.

[0014] Preferably, the bottom layer and the high layer both adopt PID controllers; the bottom layer controls each motor, and is used to adjust the angle of the ankle joint platform.

[0015] In addition, a six-dimensional force control and perception bionic foot control method is also provided, which comprises the following steps:

[0016] S1, setting a foot target posture trajectory, including expected pitch angle, expected roll angle, expected posture, expected resultant force, expected resultant torque and various expected parameters; collecting each force and torque information of the three six-dimensional force sensors in real time, and then transmitting the information to the ankle joint controller through a bus; each force is recorded as , and each torque is recorded as , i = a, q, z respectively, and correspond to the ankle joint platform, the forefoot and the hindfoot respectively;

[0017] S2, constructing a coordinate system with the ankle joint as the origin and the vertical ground upward as the z-axis, wherein the y-axis of the coordinate system is perpendicular to the center line of the forefoot six-dimensional force sensor and the hindfoot six-dimensional force sensor, and the x-axis is perpendicular to the y-axis; in the fusion layer of the ankle joint controller, each force and each corresponding torque are fused respectively to obtain and , wherein is a summation symbol, O 1 represents a reference point selected uniformly from the three six-dimensional force sensors, represents the total force synthesized by the three six-dimensional force sensors, linear represents a linear force, represent three six-dimensional force sensors to O 1 as a reference point, r represent the position vector corresponding to each six-dimensional force sensor;

[0018] Based on and , the estimated position of the zero moment point and the estimated direction of the foot posture are calculated, and a posture estimation model is constructed; the posture estimation model outputs the projection position of the zero moment point on the foot and the real-time direction of the foot posture in real time, which is used to determine the posture deviation and the gait phase;

[0019] S3, in the middle layer of the ankle joint controller, a feedback control rate is constructed according to the posture deviation, and a target control moment for adjusting the foot posture is generated; in the bottom layer, the target angles θ1 and θ2 of the two motors are solved based on the target control moment, and the two motors are controlled to adjust the posture of the ankle joint platform according to the corresponding target angles;

[0020] S4, in the high layer of the ankle joint controller, based on the gait phase, it is determined that the bionic foot main body is in the support phase or in the swing phase, and then the posture adjustment is continued according to the projection position of the zero moment point on the foot.

[0021] Preferably, after the posture adjustment is completed in the high layer, the adjusted posture in step S4 is taken as a target posture, the target posture and the gain are transmitted to the middle layer, the posture deviation is recalculated and the feedback control rate is reconstructed for feedback adjustment; wherein the gain includes the proportional gain and the differential gain of the selected parameters before and after adjustment; when the gain is transmitted, the multi-dimensional force region error , is the total force after the posture adjustment, the moment error is defined as , is the total moment after the posture adjustment, the joint feedback of the multi-dimensional force region error and the moment error is constructed as , which is used to correct the gain; , wherein, and are the basic gains under static state, and are used as adjustment coefficients, is a weight matrix of the three six-dimensional force sensors, which is used for adjustment in different directions and represents the contribution of each six-dimensional force sensor to the sensitivity of the controller.

[0022] The present application has the beneficial effects compared with the prior art:

[0023] The technical scheme of the application adopts the triangular arch support configuration with the foot arch shape in combination with the flexible bionic foot bottom material, compared with the traditional rigid flat foot structure, can realize larger area contact and adhesion on uneven ground, three six-dimensional force sensors are arranged to respectively collect the spatial force information of the forefoot, the hind foot and the ankle joint platform, and the zero moment point position, the ground normal inclination and other data are calculated through fusion to provide more sufficient feedback information for attitude control, the symmetric link driving ankle joint is adopted to replace the traditional series motor structure, not only the structure is more compact, but also the response speed is faster, the driving coupling is small, the ankle joint dynamic response and attitude adjustment capacity are enhanced, in addition, the gait stability and the terrain adaptability are improved: through the three-layer architecture introduced in the control method, combined with the real-time perceived force information and gait recognition, the control strategy can be automatically adjusted according to the support phase or the swing phase, and the whole process closed loop dynamic adjustment from impact buffering, stable support to starting and leaving the ground is realized. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 It is a whole structure schematic diagram of a six-dimensional force control and perception bionic foot system.

[0025] Figure 2 It is a structure schematic diagram of a driving system in a six-dimensional force control and perception bionic foot system.

[0026] Figure 3 It is a principle diagram of fusion calculation for estimating the contact attitude in the fusion layer.

[0027] Figure 4 It is a function flow block diagram of an ankle joint control method.

[0028] The figure mark: 1-ankle joint assembly, 2-forefoot, 3-hind foot, 4-rigid arch assembly, 5-fascia connecting piece, 6-ankle joint assembly six-dimensional force sensor, 7-forefoot six-dimensional force sensor, 8-hind foot six-dimensional force sensor, 9-motor, 10-push rod, 11-spherical joint, 12-symmetric link, 13-spherical hinge, 14-ankle joint, 15-calf rod. DETAILED DESCRIPTION

[0029] The technical scheme of the application will be described in detail below. Figures 1 to 4 The technical scheme in the embodiment of the application will be described in detail.

[0030] As shown in the figure, it is a whole structure schematic diagram of a six-dimensional force control and perception bionic foot system; as shown in the figure, it is a structure schematic diagram of a driving system in a six-dimensional force control and perception bionic foot system. Figure 1 Figure 2 Figure 1 Figure 2 ​​​As shown, in the six-dimensional force control perception bionic foot system, the entire system is constructed based on the bionic foot body, parallel driving mechanism, ankle joint platform, multi-point six-dimensional force sensing device and ankle joint controller, which can realize large-area contact and adhesion on uneven ground, and through the fusion of information of three six-dimensional force sensors, the zero moment point and other data are calculated to provide sufficient data information for attitude control. Finally, closed-loop feedback is introduced to realize dynamic adjustment, greatly enhancing the accuracy and stability of bionic foot movement.

[0031] The bionic foot body includes a foot triangular arch structure and a foot bottom, the foot bottom is used to contact the ground and support the foot triangular arch structure, and the foot triangular arch structure is used to connect the ankle joint assembly 1 in the ankle joint platform to form a stable foot architecture. The foot bottom includes a forefoot 2, a hindfoot 3 and a fascia connector 5, the fascia connector 5 is used to connect the forefoot 2 and the hindfoot 3; the forefoot 2 and the hindfoot 3 can both adopt a locust foot palm architecture, and both include a rigid connector, a locust flexible foot pad and a rigid pressing plate layer, which are connected in a sandwich structure from top to bottom, i.e. the rigid connector, the locust flexible foot pad and the rigid pressing plate layer, which are connected in a vertical direction by screws, the screws are arranged in penetration, and a plurality of screws pass through the rigid connector, the middle layer flexible foot pad layer and the rigid pressing plate layer from top to bottom, the middle part of the locust flexible foot pad layer is provided with a through hole structure for accommodating the screw rod part and preventing slipping, while ensuring the local deformation space of the flexible material. The locust flexible foot pad can be used to enhance the ground friction and ground adhesion to adapt to various different ground conditions; in the bionic foot body, the forefoot 2, the hindfoot 3 and the rigid arch assembly 4 are connected to form a triangular arch structure, which can ensure the stability of the bionic foot body.

[0032] The ankle joint platform includes the ankle joint assembly 1 and the ankle joint 14 connected thereto, and the ankle joint assembly 1 connects the rigid arch assembly 4 of the foot triangular arch structure, so that the bionic foot body and the ankle joint platform can be effectively fixed. The rigid arch assembly 4 includes a front arch segment and a rear arch segment, the front arch segment is used to connect the forefoot 2 and the ankle joint assembly 1, and the rear arch segment is used to connect the hindfoot 3 and the ankle joint assembly 1.

[0033] The multi-point six-dimensional force sensing device comprises three identical six-dimensional force sensors, which are respectively installed at the bottom of the forefoot 2, the bottom of the hindfoot 3 and the bottom of the ankle joint assembly 1, and are respectively denoted as the ankle joint assembly six-dimensional force sensor 6, the forefoot six-dimensional force sensor 7 and the hindfoot six-dimensional force sensor 8, for obtaining the force and torque information of the bionic foot body and the ankle joint platform, and then transmitting the information to the ankle joint controller through a CAN or RS485 bus. The CAN (Controller Area Network) is a serial communication network standard widely used in the fields of automobiles and industrial control, and can safely and efficiently transmit information; the RS485 bus is a serial data signal transmission standard commonly used in the fields of computers and automation, and has good anti-interference performance and long transmission distance. When each six-dimensional force sensor works, the sampling rate is not less than 100 Hz, realizing multi-point high-frequency synchronous sampling.

[0034] In the embodiment, the parallel driving mechanism comprises symmetrical links 12, two motors 9, spherical hinges 13, shank rods 15, push rods 10 and ball joints 11. The symmetrical links 12 are composed of two identical and parallel symmetrical links, and the ankle joint assembly 1 is connected to the symmetrical links 12 through the spherical hinges 13 to ensure the smoothness and response accuracy of the output; meanwhile, each link in the symmetrical links 12 is driven by the corresponding motor 9 to drive the push rod 10, and the push rod 10 and the symmetrical links 12 are connected through the ball joint 11; the two motors 9 are installed inside the shank rod 15, and the shank rod 15 is connected to the ankle joint assembly 1 through the ankle joint 14. The ankle joint 14 can adopt a universal joint, which is a mechanical connecting piece that can bear forces from different directions and has the freedom to change like the ankle during human walking, so as to realize the posture adjustment of the ankle joint 14 and adapt to various terrain conditions. In addition, the lower part of the ball joint 11 and the symmetrical links 12 are provided with threads, and the ball joint 11 and the symmetrical links 12 are connected based on the threads. Using the threaded connection can adjust the overall height of the driving symmetrical links 12 within a certain range, and the range angle of the ankle joint 14 can be adjusted according to the requirements.

[0035] Specifically, the push rod 10, the ball joint 11, the symmetrical link 12 and the spherical hinge 13 are the main mechanism for controlling the double-degree-of-freedom motion of the ankle joint 14, and the double-degree-of-freedom motion is the pitch and roll. The motor 9 controls the rotation speed and direction of the push rod 10, so that the symmetrical links 12 move up or down under the drive of the push rod 10, and the push rod 10 drives the ankle joint platform to change the angle. If the two motors 9 rotate in the same direction, the bionic foot system as a whole performs the dorsiflexion and plantarflexion motion; if the two motors 9 rotate in opposite directions, the bionic foot system performs the inversion motion.

[0036] The parallel driving mechanism can move freely in multiple directions by adopting the ball hinge 13 and the universal joint, thereby providing better flexibility and a movement mode closer to that of a real human body; each motor 9 drives the push rod 10 and is connected to the symmetrical connecting rod 12 through the ball joint 11, so that the position of each connecting rod can be controlled accurately, thereby controlling the pitch angle and the roll angle of the ankle joint, which helps to finely adjust the posture of the bionic foot; and due to the structural characteristics, the device can adapt to different ground conditions and can maintain good stability and comfort on both flat road surfaces and irregular terrains.

[0037] In the embodiment, the ankle joint controller is used to adjust the parallel driving mechanism, the parallel driving mechanism controls the ankle joint assembly 1 through the ankle joint, and further controls the bionic foot body. The ankle joint controller fuses the information transmitted by each six-dimensional force sensor, and then controls the parallel driving mechanism to adjust the ankle joint platform and the bionic foot body based on the fused result.

[0038] The ankle joint controller adopts a multi-layer control architecture, including a fusion layer, a bottom layer, a middle layer and a high layer. The fusion layer fuses each force and torque information transmitted by each six-dimensional force sensor, and is used to estimate a plurality of parameters corresponding to the movement of the bionic foot body, the plurality of parameters including total support force, total torque, contact posture and zero torque point distribution. The bottom layer and the middle layer both adopt a PID controller, the bottom layer is used to control each motor 9 in the parallel driving mechanism to adjust the angle of the ankle joint platform, and the middle layer is used to adjust the posture. The high layer is used to calculate a control strategy of the bionic foot and transmit the control strategy to the middle layer or the bottom layer to form a closed-loop feedback.

[0039] As shown in Figure 3 , it is a principle diagram for estimating the contact posture by fusion calculation in the fusion layer; as shown in Figure 4 , it is a functional flow chart of an ankle joint control method; in combination with Figure 3 and Figure 4 , the application further proposes a six-dimensional force control and perception bionic foot control method, which combines the feedback of the six-dimensional force sensors at three different point positions, posture estimation and driving execution of the symmetrical connecting rod to form a closed-loop posture control mechanism.

[0040] The method specifically includes the following steps:

[0041] S1, set the foot target posture trajectory, including the expected pitch angle, the expected roll angle, the expected posture, the expected resultant force, the expected resultant torque and a plurality of desired parameters; as shown in the first 004 step, the force and torque information of each six-dimensional force sensor is collected in real time and then transmitted to the ankle joint controller through a bus; each force is recorded as , and each torque is recorded as , i =a, q, z And respectively corresponding to the ankle joint platform, the forefoot 2, the hindfoot 3.

[0042] S2, taking the ankle joint 14 as the origin and the vertical ground upward as the z-axis to construct the coordinate system, wherein the y-axis of the coordinate system is vertically directed to the center line of the forefoot six-axis force sensor 7 and the hindfoot six-axis force sensor 8 from the origin, and the x-axis is perpendicular to the y-axis; in the fusion layer of the ankle joint controller, as shown in steps 001 and 005, each force and each torque corresponding to each force is fused in the fusion layer respectively, to obtain and ; wherein, is the summation symbol, O 1 represents a reference point selected uniformly by the three six-axis force sensors, which also corresponds to the center point of the ankle joint platform, represents the total force synthesized by the three six-axis force sensors, linear represents the linear force relative to the torque, used to distinguish whether it belongs to linear information or rotational information, representing the change of the direction of motion; represents the total torque synthesized by the three six-axis force sensors with O 1 as the reference point, r represents the position vector corresponding to each six-axis force sensor, that is, the direction vector of each six-axis force sensor relative to the reference point O 1.

[0043] Then, as shown in step 002, based on and , the current estimated parameters such as the attitude deviation, the estimated position of the zero moment point and the estimated direction of the foot posture can be estimated, and the attitude estimation model is constructed. As shown in step 003, the attitude estimation model based on and , the projection position of the zero moment point on the foot, the support area, the ground normal inclination and the like are output in real time, and the ground normal inclination can also reflect the real-time direction of the foot posture; and based on the three-point space torque balance equation which has been widely applied at present, the total force state can be calculated, and combined with the parameters such as the ground normal inclination and the estimated direction which have been calculated, the gait phase can be effectively identified.

[0044] The core objective of the pose estimation model is to calculate the spatial pose of the current foot bottom contact, such as direction, deviation, and support shape, based on the three-dimensional arrangement of the bionic foot bottom, the measurement information of the six-dimensional force sensor, and the subsequent control and adjustment to provide input. The construction process of the model is summarized as follows: 1) input parameters, the position coordinates of each six-dimensional force sensor, the corresponding force and torque data; 2) calculate the total force and torque; 3) calculate the zero moment point, calculate the projection coordinates of the zero moment point on the foot bottom plane; 4) calculate the normal of the foot bottom contact surface, consider the force direction of each six-dimensional force sensor as the measurement response to the normal of the ground, calculate the unit force direction of each six-dimensional force sensor , then calculate the weighted average to obtain the estimated normal of the foot bottom contact surface n est , ; 5) calculate the pose deviation angle, with the vertical reference normal n ref =[0,0,1] T , calculate the pose deviation angle θ err =arccos(n est ⋅n ref ), arccos represents the inverse cosine function; 6) estimate the support area, combine the force distribution of the three six-dimensional force sensors and the mechanical model of the flexible material, and indirectly estimate the effective contact area of the foot bottom by comprehensive analysis of the pressure distribution and foot bottom deformation, set the effective pressure action area of each six-dimensional force sensor, assume that the force affected radius is r i , combine the three circular influence areas in the two-dimensional plane, and calculate the combined area as the estimation result of the contact area.

[0045] S3, as shown in steps 006, 007, 008 and 009, in the middle layer of the ankle joint controller, according to the pose deviation between the current pose and the expected pose, a feedback control rate is constructed to generate a target control torque for adjusting the pose, combined with the inverse kinematics calculation of the parallel driving mechanism, the target output torque is mapped to the angle parameter required by each motor 9 to output . Subsequently, under the control of the bottom layer, each motor 9 drives the symmetrical connecting rod to adjust the pose of the ankle joint platform according to the corresponding target angle, and realizes the pose adjustment of the ankle joint platform.

[0046] S4, at the high layer of the ankle joint controller, based on the gait phase, it is determined that the bionic foot subject is in the support phase or in the swing phase, and then the posture adjustment is continued according to the projection position of the zero moment point on the foot bottom. After the posture adjustment is completed at the high layer, as shown in steps 009 and 010, the control process can also automatically adjust the control parameters according to the support phase, swing phase and other data, transmit the adjusted target posture required angles and calculated gains to the middle layer, recalculate the posture deviation and reconstruct the feedback control rate, that is, return to step 006 to constitute feedback adjustment, and realize the rapid response and stable support to different terrains and dynamic disturbances.

[0047] Through the three-layer architecture introduced in the control method, the bottom layer controls two motors 9, the middle layer adjusts the posture, and the high layer judges the closed-loop strategy. Combined with the real-time sensing of the force information and the gait phase recognition, the control strategy can be automatically adjusted according to the support phase or the swing phase, the whole process dynamic adjustment from impact buffering, stable support to starting and leaving the ground is realized, and the gait stability and terrain adaptability are further improved.

[0048] Specifically, the gain includes the proportional gain and the differential gain before and after the selected parameters are adjusted; when the transmission gain is transmitted, the multi-dimensional force region error is first constructed, is the total force after the posture is adjusted, that is, the expected total force, the moment error is defined as , is the total moment after the posture is adjusted, that is, the expected total moment, the joint feedback of the multi-dimensional force region error and the moment error is constructed as , which is used to modify the gain; , wherein, and are the basic gains measured by the bionic foot in the static state, and are used as adjustment coefficients, is a weight matrix set by the three six-dimensional force sensors, which is used for adjustment in different directions and represents the contribution of each six-dimensional force sensor to the sensitivity of the controller.

[0049] In summary, the present application adopts the triangular arch support configuration with the shape of the arch of the foot and combines with the flexible biomimetic foot bottom material, compared with the traditional rigid flat foot structure, can realize larger area contact and fit on uneven ground; three six-dimensional force sensors are arranged, respectively collecting the spatial force information of the forefoot 2, the hindfoot 3 and the ankle joint platform, and through the three-point moment model fusion, the zero moment point position, the ground normal inclination and other data are calculated, providing more sufficient feedback information for attitude control; the symmetric connecting rod is used to drive the ankle joint 14, replacing the traditional series motor 9 structure, not only the structure is more compact, but also the response speed is faster, the driving coupling is small, the dynamic response and attitude adjustment ability of the ankle joint 14 are enhanced; in addition, the gait stability and the terrain adaptability are improved: through the three-layer architecture introduced in the control method, combined with the real-time sensing force information and gait recognition, the control strategy can be automatically adjusted according to the support phase or the swing phase, realizing the whole process closed-loop dynamic adjustment from impact buffering, stable support to starting and leaving the ground, which has significant progressiveness.

[0050] The above examples only illustrate the technical idea of the present application, and cannot limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical scheme falls within the protection scope of the present application.

Claims

1. A six-dimensional force controlled sensory bionic foot system, characterized in that, The bionic foot body, the parallel driving mechanism, the ankle joint platform, the multi-point six-dimensional force sensor device and the ankle joint controller, the multi-point six-dimensional force sensor device is used to obtain the information of the bionic foot body and the ankle joint platform, the ankle joint controller is used to adjust the parallel driving mechanism, and the parallel driving mechanism controls the bionic foot body through the ankle joint platform; Wherein, the bionic foot body includes a foot triangular arch structure and a foot bottom, the foot bottom is used to contact the ground and support the foot triangular arch structure; the foot bottom includes a forefoot (2), a hindfoot (3) and a fascia connector (5), the fascia connector (5) is used to connect the forefoot (2) and the hindfoot (3); the forefoot (2) and the hindfoot (3) both adopt a locust foot palm architecture, each locust foot palm architecture includes a rigid connector, a locust flexible foot pad and a rigid pressing plate layer connected in a sandwich structure, and each locust foot palm architecture is vertically penetrated by a screw from top to bottom through the rigid connector, the locust flexible foot pad and the rigid pressing plate layer, and the middle part of the locust flexible foot pad layer is provided with a through hole structure for accommodating the screw; The ankle joint platform includes an ankle joint (14) and an ankle joint assembly (1) connected, and the ankle joint assembly (1) is connected with the foot triangular arch structure; the multi-point six-dimensional force sensor device includes an ankle joint assembly six-dimensional force sensor (6), and further includes two six-dimensional force sensors arranged on the forefoot (2) and the hindfoot (3), each six-dimensional force sensor is used to collect each force and torque, and then the information is transmitted to the ankle joint controller through a bus, each force is recorded as and each torque is recorded as, i=a, q, z and corresponds to the ankle joint platform, the forefoot (2) and the hindfoot (3) respectively; The ankle joint controller adopts a multi-layer control architecture, including a fusion layer, a bottom layer, a middle layer and a high layer, the ankle joint controller fuses the information transmitted by each six-dimensional force sensor, and then controls the parallel driving mechanism to adjust the ankle joint platform and the bionic foot body based on the fusion result, and each motor (9) is arranged in the parallel driving mechanism; the bottom layer solves the target angles θ1 and θ2 of the two motors (9) based on the target control torque, and then controls the two motors (9) to adjust the posture of the ankle joint platform according to the corresponding target angles; the middle layer adopts a posture correction controller, the posture correction controller constructs a feedback control rate according to the posture deviation, and generates a target control torque for adjusting the posture of the foot bottom; the high layer is used to calculate the control strategy of the bionic foot and transmit the control strategy to the middle layer or the bottom layer; the bottom layer and the high layer both adopt a PID controller; the fusion layer fuses each force and torque transmitted by each six-dimensional force sensor, and is used to estimate a plurality of parameters corresponding to the movement of the bionic foot body, the plurality of parameters including total support force, total torque, contact posture and zero torque point distribution. Wherein, when fusion is carried out in the fusion layer, a coordinate system is constructed with the ankle joint (14) as the origin and the vertical ground upward as the z axis, the y axis of the coordinate system is vertically directed to the center line of the forefoot six-dimensional force sensor (7) and the rearfoot six-dimensional force sensor (8) from the origin, and the x axis is perpendicular to the y axis; then each force and each torque corresponding to each force are fused respectively to obtain and , is a summation symbol, O 1 represents a reference point selected uniformly by the three six-dimensional force sensors, represents the total force synthesized by the three six-dimensional force sensors, linear represents a linear force, represents the total torque synthesized by the three six-dimensional force sensors with O 1 as the reference point, r represents a position vector corresponding to each six-dimensional force sensor; and are used for calculating the estimated position of the zero moment point and the estimated direction of the foot posture, constructing a posture estimation model; the posture estimation model is used for real-time outputting the projection position of the zero moment point on the foot and the real-time direction of the foot posture, so as to determine the posture deviation and the gait phase. When the higher-level ankle joint controller performs calculations, it first determines whether the bionic foot is in the support or swing phase based on the gait phase. Then, it continues to adjust the posture based on the projection position of the zero-moment point on the sole of the foot. After the higher-level controller completes the posture adjustment, it uses the adjusted posture as the target posture and transmits the target posture and gain to the middle level. The posture deviation is recalculated and the feedback control law is reconstructed to form the feedback regulation. The gain includes the proportional gain of the selected parameter before and after the adjustment. and differential gain When calculating transmission gain, first construct a multidimensional force region error. , It is the total force after attitude adjustment, and the torque error is defined as... , It is the total torque after attitude adjustment, constructing a joint feedback of multi-dimensional force region error and torque error. Used to correct gain; , ,in, and These are all base gains under static conditions. and All are used as adjustment coefficients. A weighting matrix is ​​set for the three six-dimensional force sensors to adjust in different directions and characterize the contribution of each six-dimensional force sensor to the controller sensitivity.

2. The six-axis force-controllable sensory bionic foot system according to claim 1, wherein, The foot triangular arch structure adopts a rigid arch assembly (4) for fixing the bionic foot body and the ankle joint platform; the rigid arch assembly (4) includes a front arch segment and a rear arch segment, the front arch segment is used to connect the forefoot (2) and the ankle joint assembly (1), and the rear arch segment is used to connect the hindfoot (3) and the ankle joint assembly (1).

3. The six-axis force-controllable sensory bionic foot system according to claim 1, wherein, Parallel driving mechanism includes symmetrical connecting rod (12), two motors (9), spherical hinge (13), calf rod (15), push rod (10), spherical joint (11); ankle joint assembly (1) is connected with symmetrical connecting rod (12) through spherical hinge (13); symmetrical connecting rod (12) drives push rod (10) based on two motors (9), push rod (10) and symmetrical connecting rod (12) are connected through spherical joint (11); two motors (9) are installed inside calf rod (15), calf rod (15) is connected with ankle joint assembly (1) through ankle joint (14).

4. The six-axis force-controllable sensory bionic foot system according to claim 3, characterized in that, The lower part of the spherical joint (11) and the symmetrical connecting rod (12) are provided with threads, and the spherical joint (11) and the symmetrical connecting rod (12) are connected based on the threads.

5. A control method of a six-axis force-controllable and perception bionic foot, which is operated by using the six-axis force-controllable and perception bionic foot system according to any one of claims 1 to 4, characterized in that, The method comprises the following steps: S1, set the foot target posture trajectory, including the expected pitch angle, the expected roll angle, the expected posture, the expected resultant force, the expected resultant moment and other desired parameters; real-time collection of each force and moment information of the three six-dimensional force sensors, and then transmission to the ankle joint controller through the bus; each force is recorded as , and each moment is recorded as , i = a, q, z , and the ankle joint platform, the forefoot (2) and the hindfoot (3) correspond respectively. S2, a coordinate system is constructed with the ankle joint (14) as the origin and the z-axis upward perpendicular to the ground, wherein the y-axis of the coordinate system is directed vertically from the origin to the center line of the forefoot six-axis force sensor (7) and the rearfoot six-axis force sensor (8), and the x-axis is perpendicular to the y-axis; in the fusion layer of the ankle joint controller, each force and each torque corresponding thereto is fused respectively to obtain and wherein, is a summation symbol, O 1 represents a reference point selected uniformly by the three six-axis force sensors, represents the total force synthesized by the three six-axis force sensors, linear represents the linear force, represents the total torque synthesized by the three six-axis force sensors with O 1 as the reference point, r represents the position vector corresponding to each six-axis force sensor; Based on and , the estimated position of the zero moment point and the estimated direction of the foot posture are calculated, and a posture estimation model is constructed; the posture estimation model outputs the projection position of the zero moment point on the foot and the real-time direction of the foot posture in real time, which are used to determine the posture deviation and the gait phase; S3, in the middle layer of the ankle joint controller, the feedback control rate is constructed according to the posture deviation, and the target control torque for adjusting the posture of the sole is generated; in the bottom layer, the target angles θ1 and θ2 of the two motors (9) are solved based on the target control torque, and then the two motors (9) are controlled to adjust the posture of the ankle joint platform according to the corresponding target angles; S4, in the high layer of the ankle joint controller, based on the gait phase, it is judged that the bionic foot main body is in the support phase or in the swing phase, and then the posture adjustment is continued according to the projection position of the zero moment point on the sole.

6. The control method of a six-axis force-controllable bionic foot according to claim 5, characterized in that, After completing the posture adjustment in the high layer, the adjusted posture in step S4 is taken as the target posture, the target posture and the gain are transmitted to the middle layer, the posture deviation is recalculated and the feedback control rate is reconstructed for feedback adjustment. wherein the gains include proportional gains of the selected parameters before and after adjustment and derivative gains ; when the transmission gain is constructed, the multi-dimensional force region error is constructed first , is the total force after adjusting the posture, the moment error is defined as , is the total moment after adjusting the posture, the joint feedback of the multi-dimensional force region error and the moment error is constructed as , which is used to correct the gain; , wherein, and are the basic gains under static state, and are used as adjustment coefficients, is the weight matrix set for the three six-dimensional force sensors, which is used for adjustment in different directions and represents the contribution of each six-dimensional force sensor to the sensitivity of the controller.

Images