CN117930663B - Motion control system of four-foot robot based on eight-element neural network - Google Patents
Motion control system of four-foot robot based on eight-element neural network Download PDFInfo
- Publication number
- CN117930663B CN117930663B CN202410320353.3A CN202410320353A CN117930663B CN 117930663 B CN117930663 B CN 117930663B CN 202410320353 A CN202410320353 A CN 202410320353A CN 117930663 B CN117930663 B CN 117930663B
- Authority
- CN
- China
- Prior art keywords
- neuron
- neurons
- signal
- gait
- unidirectional coupling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 61
- 230000033001 locomotion Effects 0.000 title claims abstract description 40
- 210000002569 neuron Anatomy 0.000 claims abstract description 288
- 230000005021 gait Effects 0.000 claims abstract description 79
- 210000004394 hip joint Anatomy 0.000 claims abstract description 28
- 210000000629 knee joint Anatomy 0.000 claims abstract description 28
- 230000033764 rhythmic process Effects 0.000 claims abstract description 28
- 230000033228 biological regulation Effects 0.000 claims abstract description 19
- 210000002414 leg Anatomy 0.000 claims abstract description 17
- 238000012805 post-processing Methods 0.000 claims abstract description 10
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 claims abstract description 9
- 238000006073 displacement reaction Methods 0.000 claims abstract description 4
- 238000010168 coupling process Methods 0.000 claims description 110
- 238000005859 coupling reaction Methods 0.000 claims description 110
- 230000008878 coupling Effects 0.000 claims description 109
- 238000000034 method Methods 0.000 claims description 22
- 230000005764 inhibitory process Effects 0.000 claims description 14
- 230000002401 inhibitory effect Effects 0.000 claims description 12
- 230000001808 coupling effect Effects 0.000 claims description 10
- 230000002457 bidirectional effect Effects 0.000 claims description 9
- 210000001503 joint Anatomy 0.000 claims description 8
- 210000001364 upper extremity Anatomy 0.000 claims description 8
- 230000009184 walking Effects 0.000 claims description 8
- 230000009191 jumping Effects 0.000 claims description 7
- 239000012528 membrane Substances 0.000 claims description 7
- 238000013461 design Methods 0.000 abstract description 13
- 230000007613 environmental effect Effects 0.000 abstract description 7
- 230000037230 mobility Effects 0.000 abstract description 2
- 230000001276 controlling effect Effects 0.000 description 21
- 238000004364 calculation method Methods 0.000 description 7
- 210000001624 hip Anatomy 0.000 description 7
- 210000003127 knee Anatomy 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 239000000243 solution Substances 0.000 description 6
- 238000004088 simulation Methods 0.000 description 5
- 230000003750 conditioning effect Effects 0.000 description 3
- 125000004122 cyclic group Chemical group 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000001537 neural effect Effects 0.000 description 3
- 239000011664 nicotinic acid Substances 0.000 description 3
- 238000010587 phase diagram Methods 0.000 description 3
- 230000001953 sensory effect Effects 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000002347 injection Methods 0.000 description 2
- 239000007924 injection Substances 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 230000002787 reinforcement Effects 0.000 description 2
- 230000001020 rhythmical effect Effects 0.000 description 2
- 241000269333 Caudata Species 0.000 description 1
- 241000238631 Hexapoda Species 0.000 description 1
- 241000282373 Panthera pardus Species 0.000 description 1
- 241000251539 Vertebrata <Metazoa> Species 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000003169 central nervous system Anatomy 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- RKTYLMNFRDHKIL-UHFFFAOYSA-N copper;5,10,15,20-tetraphenylporphyrin-22,24-diide Chemical compound [Cu+2].C1=CC(C(=C2C=CC([N-]2)=C(C=2C=CC=CC=2)C=2C=CC(N=2)=C(C=2C=CC=CC=2)C2=CC=C3[N-]2)C=2C=CC=CC=2)=NC1=C3C1=CC=CC=C1 RKTYLMNFRDHKIL-UHFFFAOYSA-N 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 210000000609 ganglia Anatomy 0.000 description 1
- 210000003016 hypothalamus Anatomy 0.000 description 1
- 230000003137 locomotive effect Effects 0.000 description 1
- 210000001259 mesencephalon Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 230000027387 rhythmic behavior Effects 0.000 description 1
- 229910001415 sodium ion Inorganic materials 0.000 description 1
- 238000013112 stability test Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009182 swimming Effects 0.000 description 1
- 210000000225 synapse Anatomy 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a motion control system of a four-foot robot based on an eight-element neural network, and belongs to the field of motion control of four-foot robots. The system comprises: the signal regulation and control module is used for generating gait control parameters; the signal generation module is used for receiving the gait control parameters generated by the signal regulation and control module, inputting the gait control parameters into the eight-element neural network, and respectively solving ordinary differential equations of the eight neurons to obtain gait rhythm signals; the signal post-processing module is used for receiving the gait rhythm signal obtained by solving in the signal generating module and converting the gait rhythm signal into a displacement signal which correspondingly controls eight joints on four legs of the four-legged robot to actuate. The invention designs the rhythmicity of the gait of the central pattern generator and the phase relation of the hip and knee joints based on the symmetry principle, realizes the low-calculation-force and high-reliability control of the multi-joints of the quadruped robot, and improves the mobility and environmental adaptability of the quadruped robot under 5 gaits.
Description
Technical Field
The invention belongs to the field of motion control of four-foot robots, and particularly relates to a motion control system of a four-foot robot based on an eight-element neural network.
Background
In recent years, compared with a wheeled robot, the legged robot has better maneuverability, terrain adaptability and stability in an unstructured environment, and is suitable for complex tasks such as military exploration, disaster search and rescue and the like. The four-legged robot shows better maneuverability and stability in all legged robots. The multifunctional walking device not only can flexibly execute tasks in various complex terrain environments, but also can stably walk under unstructured terrains such as mud, grasslands and the like.
At present, common methods in the field of four-legged robot control are model-based control, learning-based control, and central pattern generator control, respectively. The model-based control method is a classical control method in the field of robots, and is commonly used for accurately controlling large quadruped robots, such as leopard robots, big Dog robot and the like. The control method needs to accurately model and plan the motion of the robot model and the environment, and has the defects of complex calculation, poor environmental adaptability and the like. Learning-based control is a method for designing and improving a control strategy by using reinforcement learning and the like, and can realize accurate control of a robot in a complex, nonlinear or uncertain environment, but the method is highly dependent on data and training environment, and still faces great challenges when a model is transferred to a real environment.
A central pattern Generator (CENTRAL PATTERN Generator, CPG) is a small neural network formed by a series of interacting neuronal models or oscillators coupled, which has been widely demonstrated to exist in the central nervous system of vertebrates and in the ganglia of invertebrates. CPG can generate a fundamental signal of rhythmic behavior (e.g., respiration or motion) without sensory feedback, but requires sensory feedback to regulate CPG signals. In the field of robot control, CPG is widely used to control the movements and behaviors of various robotic systems. For example: the motion control method is used for researching the salamander robot for swimming and walking, controlling the motion of the bipedal quadruped and hexapod robots combined with sensory feedback, controlling the motion of the electronic pneumatic quadruped robot, and controlling the motion of the quadruped robot, the soft snake-shaped robot and the like combined with reinforcement learning.
Compared with the other two methods, the CPG has the greatest advantages of simple structure and high calculation efficiency. The coordinated rhythmic motion control of the plurality of joints of the robot can be realized by only calculating a normal differential equation set of the power system. Furthermore, CPG has inherent stability and adaptivity.
Currently, most four-foot motion controlled CPG networks are typically four-neuron networks. The device has simple structure, is easy to integrate into a motion controller, and has good compatibility with sensors and other learning algorithms. It also has some drawbacks, firstly, the number of gait rhythms that can be generated is limited. Common gait for quadruped animals includes: walking (walk), jogging (stro), running (pace), jumping (bound), leaping (pronk), jumping (jump), and the like, and the phase relationship is shown as a-f in fig. 1. Most four-neuron networks can only achieve no more than three gait types (typically walking, jogging, jumping). Second, the number of joints that can be controlled is limited (typically four). The most common design of current quadruped robots is three joints per leg: hip abduction-adduction, hip flexion-extension and knee flexion-extension. However, in CPGs of all four neurons, even eight neurons, at most only four signals can be generated for controlling the phase relationship between the four legs. A mapping method is generally needed to map the signals of a single neuron into two joint position signals to achieve a phase relationship between the knee joint and the hip joint. From a biological point of view, signals controlling multiple joints are likely to be generated by specific neurons, which requires a more complex network architecture.
Designing a network architecture that is more complex than a four-neuron network does not impair its advantages, and a network with additional neurons may also contain more symmetry, thereby generating more gait types. Furthermore, the mechanism of hip-knee joint coordination control through the inherent characteristics of the network can deepen understanding of gait control mechanisms of quadruped animals, which also helps to promote related studies of robotics and biology.
One possible way to design a CPG network is the symmetry principle. If the network is symmetrical, constraints imposed on dynamics by symmetry typically result in neuronal synchronization or phase locking. The rhythmic nature of gait is here a kind of time-space symmetry of the time-domain signal between the joints. Symmetry of the network structure may constrain symmetry of the CPG neuron signals. However, how to design the rhythmicity of CPG gait and the phase relation of hip and knee joints specifically, low-calculation force and high-mobility control of the multi-joints of the quadruped robot is realized, and multiple gaits are further realized, so that mobility and environmental adaptability of a robot system are improved, and an efficient solution is still lacking in the prior art.
Disclosure of Invention
The invention aims to solve the problem that the four-foot robot in the prior art is limited in the number of degrees of freedom and types of rhythm signals, and provides a four-foot robot motion control system based on an eight-element neural network, so that eight degrees of freedom coordination control on hip joints and knee joints of the four-foot robot is realized, five gait rhythm signals can be generated, and the environmental adaptability of the system can be improved.
The specific technical scheme adopted by the invention is as follows:
A four-foot robot motion control system based on an eight-element neural network, comprising:
The signal regulation and control module is used for generating gait control parameters;
The signal generation module is used for receiving the gait control parameters generated by the signal regulation and control module, inputting the gait control parameters into the eight-element neural network, and respectively solving ordinary differential equations of the eight neurons to obtain gait rhythm signals; the eight-element neural network consists of a first unidirectional coupling network layer and a second unidirectional coupling network layer which have quadruple rotational symmetry respectively; the first unidirectional coupling network layer is formed by sequentially connecting a first neuron, a third neuron, a second neuron and a fourth neuron end to form a unidirectional coupling annular network, the second unidirectional coupling network layer is formed by sequentially connecting a fifth neuron, an eighth neuron, a sixth neuron and a seventh neuron end to form a unidirectional coupling annular network, and the coupling direction of the neurons in the first unidirectional coupling network layer is opposite to the coupling direction of the neurons in the second unidirectional coupling network layer; two-by-two neurons are used as a group between the first unidirectional coupling network layer and the second unidirectional coupling network layer, bidirectional coupling is formed between the two unidirectional coupling network layers, and four groups of neurons respectively control four legs of the four-legged robot in a one-to-one correspondence manner; and in each group of neurons, the neurons in the first unidirectional coupling network layer are used for controlling the hip joint, and the neurons in the second unidirectional coupling network layer are used for controlling the knee joint; in the ordinary differential equation corresponding to the eight-element neural network, the driving signal of each neuron is determined by three parts, namely gait control parameters generated by a signal regulation and control module, the coupling effect of the neurons from the same unidirectional coupling network layer and the coupling effect of the neurons from another unidirectional coupling network layer;
the signal post-processing module is used for receiving the gait rhythm signal obtained by solving in the signal generating module and converting the gait rhythm signal into a displacement signal which correspondingly controls eight joints on four legs of the four-legged robot to actuate.
Preferably, the gait pattern controllable by the quadruped robot comprises walking, jogging, running, jumping and leaping.
Preferably, in the eight-element neural network, the first neuron and the fifth neuron are a group for controlling the hip joint and the knee joint of the left rear leg, the second neuron and the sixth neuron are a group for controlling the hip joint and the knee joint of the right rear leg, the third neuron and the seventh neuron are a group for controlling the hip joint and the knee joint of the right front leg, and the fourth neuron and the eighth neuron are a group for controlling the hip joint and the knee joint of the left front leg.
Preferably, in the eight-element neural network, the coupling existing between the neurons adopts inhibitory coupling. Further, wherein the first neuron inhibits a third neuron, the third neuron inhibits a second neuron, the second neuron inhibits a fourth neuron, the fourth neuron inhibits a first neuron, the fifth neuron inhibits an eighth neuron, the eighth neuron inhibits a sixth neuron, the sixth neuron inhibits a seventh neuron, the seventh neuron inhibits a fifth neuron, bidirectional inhibition is provided between the first neuron and the fifth neuron, bidirectional inhibition is provided between the second neuron and the sixth neuron, bidirectional inhibition is provided between the third neuron and the seventh neuron, and the fourth neuron and the eighth neuron are bidirectional inhibition is provided between a group.
Preferably, in the eight-element neural network, a Stein neuron model is adopted as a normal differential equation of any one of the neurons i, wherein a driving signal of any one of the neurons i isWherein、、Are gait control parameters generated by the signal regulation and control module, t is time,AndThe method comprises the steps of respectively obtaining the same-layer weight super parameter and the different-layer weight super parameter, wherein X is the membrane potential sum of neurons which form inhibition on the neurons i in a unidirectional coupling network layer where the current neurons i are located, and Y is the membrane potential sum of neurons which form inhibition on the neurons i in the unidirectional coupling network layer without the current neurons i.
Preferably, when the ordinary differential equation is solved, the same-layer weight super-parameters of ordinary differential equations corresponding to eight neurons are all set to be the same, and the different-layer weight super-parameters of ordinary differential equations corresponding to neurons in the second unidirectional coupling network layer are larger than the different-layer weight super-parameters of ordinary differential equations corresponding to neurons in the first unidirectional coupling network layer in absolute value.
Preferably, the same-layer weight hyper-parameters of the ordinary differential equations corresponding to the eight neurons are all set to-0.15.
Preferably, the different layer weight super parameters of the ordinary differential equation corresponding to the neurons in the first unidirectional coupling network layer are all set to-0.1, and the different layer weight super parameters of the ordinary differential equation corresponding to the neurons in the second unidirectional coupling network layer are all set to-0.6.
Preferably, the ordinary differential equation is solved numerically by a fourth-order Dragon-Gerdostane method.
Preferably, the gait rhythm signal of each neuron solved by the signal generating module is a time domain signal with a value range between 0 and 1, and the signal post-processing module converts the time domain signal into a joint position signal or a voltage signal for controlling the corresponding joint to actuate.
Compared with the prior art, the invention has the following beneficial effects:
1. Compared with a common CPG quaternary network architecture, the network architecture provided by the invention can control more degrees of freedom of the four-legged robot, realize coordinated control over hip and knee joints, and can maintain phase locking between the hip and knee joints.
2. The network architecture of the invention can generate more gait rhythm signal types, so that the quadruped robot has better environmental adaptability.
3. Compared with the calculation scale and calculation power required by a model-based method or a learning-based method, the bionic motion control architecture based on the CPG octant D 4 symmetrical network provided by the invention has high calculation efficiency and is simpler, and only parameters of a signal regulation and control layer are required to be set, so that a set of ordinary differential equations are subjected to numerical solution, and the rhythm signal of gait can be obtained.
Drawings
FIG. 1 is a schematic diagram of gait phase relationship of a quadruped robot;
FIG. 2 is a schematic diagram of a three-layer bionic motion control architecture employed in the present invention;
FIG. 3 is a schematic diagram of an eight-element neural network architecture employed in the present invention;
Fig. 4 shows simulation results in an example of the present invention, wherein (a) - (e) are phase diagram numerical simulation results of two-layer network output state quantities and hip-knee neurons in five gait, and (f) is a stability test result after disturbance injection.
Detailed Description
In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to the appended drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The present invention may be embodied in many other forms than described herein and similarly modified by those skilled in the art without departing from the spirit of the invention, whereby the invention is not limited to the specific embodiments disclosed below. The technical features of the embodiments of the invention can be combined correspondingly on the premise of no mutual conflict.
In the description of the present invention, it will be understood that when an element is referred to as being "connected" to another element, it can be directly connected to the other element or be indirectly connected with intervening elements present. In contrast, when an element is referred to as being "directly connected" to another element, there are no intervening elements present.
In the description of the present invention, it should be understood that the terms "first" and "second" are used solely for the purpose of distinguishing between the descriptions and not necessarily for the purpose of indicating or implying a relative importance or implicitly indicating the number of features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature.
In a preferred embodiment of the present invention, an eight-element neural network-based motion control system for a quadruped robot is provided, which can generate different gait rhythm signals and maintain phase locking of hip and knee joints by constructing an eight-element Central Pattern Generator (CPG) network, so as to realize low-calculation-force and high-mobility control for the quadruped robot. In addition, the common quaternary network generally generates three gait rhythm signals, and the eight-element network consisting of eight neurons (namely the first neuron to the eighth neuron) is designed based on a symmetry principle, so that five gait rhythm signals can be generated, and the environmental adaptability of the system can be improved.
As shown in fig. 2, the motion control system of the quadruped robot adopts three layers of bionic motion control architecture, namely a signal regulation layer, a signal generation layer and a signal post-processing layer, which can be realized through corresponding functional modules, and the three layers of architecture are described in detail below.
The first layer control framework is a signal regulation module and is used for generating gait control parameters. In an embodiment of the present invention, the signal conditioning module may condition the subsequent driving signal by generating control parameters for conditioning CPG in a manner that the midbrain motor region (MESENCEPHALIC LOCOMOTOR REGION, MLR for short) of the biological hypothalamus generates the gait command signalTo adjust the gait rhythm signal of the CPG network.
The second-layer control architecture is a signal generation module and is used for receiving gait control parameters generated by the signal regulation and control module, inputting the gait control parameters into the eight-neuron neural network, and respectively solving ordinary differential equations of the eight neurons to obtain gait rhythm signals.
The third layer of control architecture is a signal post-processing module and is used for receiving gait rhythm signals obtained by solving in the signal generating module and converting the gait rhythm signals into displacement signals which correspondingly control eight joints on four legs of the four-legged robot to actuate.
In the embodiment of the invention, eight Stein neurons can be selected to construct a CPG eight-element symmetrical neural network, namely, a neural loop serving as a living organism, and modulation parameters from a signal regulation and control module are received to generate rhythm signals of different gaits and gait switching. Because the eight-element neural network is a core for realizing the motion control of the four-legged robot, in order to facilitate understanding of the principle and advantages of the network, the design idea of the eight-element neural network is described in detail through two design links. In addition, for convenience of the following description, the first neuron, the second neuron, the third neuron, the fourth neuron, the fifth neuron, the sixth neuron, the seventh neuron and the eighth neuron in the eight-element neural network are named as neuron 1, neuron 2, neuron 3, neuron 4, neuron 5, neuron 6, neuron 7 and neuron 8 respectively, but it is understood that the numbering and naming manners of the neurons do not substantially affect the implementation effect of the technical scheme.
Design element 1, global symmetry for designing gait rhythms
Because the invention needs to design a CPG network architecture for motion control of the quadruped robot, the CPG network architecture can realize walking (walk), jogging (trot), running (pace), jumping (bound) and soaring (pronk) gait. In networks with Z 4 symmetry (i.e., quad cyclic symmetry, also known as quad cyclic symmetry), the spatiotemporal symmetry of the jogging and running gait is always conjugated, meaning that the jogging and running gait cannot coexist in a single Z 4 network. The invention therefore designs the global symmetry of the network as D 4, which is a four-way network with bi-directional coupling, indicating that the network has a fourfold rotational symmetry (also known as tetragonal symmetry or square symmetry). In this network, four neurons 1, 2, 3 and 4 are used to control the left rear Leg (LH), right rear leg (RH), right front leg (RF) and left front Leg (LF) of a four-legged robot, respectively. The generator elements of the D 4 group are ω= (1324) and κ= (13) (24). The group element of D 4 can be expressed as:
The cyclic quotient subgroup and the isotropic subgroup of five gait according to the need of the invention are found in the symmetric subgroup D 4 of the bi-directionally coupled quaternary network described above. Table 1 summarizes the subgroup selection for all gaits, x i representing the states of the neurons, listing the phase relationship of other neurons relative to neuron No. 1 at different gaits. It was thus demonstrated that a bi-directional coupled four-neuron network with D 4 symmetry meets the spatiotemporal symmetry requirements of all desired gait. The global symmetry of the final eight-element CPG network architecture to be designed is set to D 4 symmetry.
Table 1 subgroup selection for all gait
Design link 2, design local symmetry of hip-knee phase lock
In order to realize the coordination control of the hip and knee joints of the quadruped robot and maintain the corresponding phase relation, two problems are required to be further solved in the design link 2 on the basis of the design link 1:
First, the bi-directional coupling quaternary network designed in the above link is extended to an eight-element network, and the D 4 symmetry of the network must be maintained.
Second, the local symmetry of the hip-knee neurons is designed to achieve a locked phase between the hip-knee.
Thus, based on the bi-directional coupled quaternary D 4 network described above, a single neuron is split into a small group of two neurons to increase the number of neurons in the network to 8. Within each group of neurons, two neurons establish a bi-directional connection, ensuring global symmetry while ensuring local symmetry as Z 2. In the extended octant network, the generator corresponding to the local symmetry is λ= (15) (26) (37) (48), so that the generator of the octant network is obtained as follows:
Finally, the eight-neuron network obtained by expanding the invention is shown in fig. 3. The eight-element neural network is divided into an upper layer and a lower layer, wherein the upper layer of neurons are used for controlling the hip joint and are called a first unidirectional coupling network layer, and the lower layer of neurons are used for controlling the knee joint and are called a second unidirectional coupling network layer. The first unidirectional coupling network layer of the upper layer of the network and the second unidirectional coupling network layer of the lower layer of the network both have D 4 symmetry. The first unidirectional coupling network layer is formed by sequentially connecting a first neuron, a third neuron, a second neuron and a fourth neuron end to form a unidirectional coupling annular network, namely the first neuron points to the third neuron, the third neuron points to the second neuron, the second neuron points to the fourth neuron, and the fourth neuron points to the first neuron. Likewise, the second unidirectional coupling network layer is formed by sequentially connecting a fifth neuron, an eighth neuron, a sixth neuron and a seventh neuron end to form a unidirectional coupling annular network, namely the fifth neuron points to the eighth neuron, the eighth neuron points to the sixth neuron, the sixth neuron points to the seventh neuron and the seventh neuron points to the fifth neuron. Thus, the coupling direction of the neurons in the first unidirectional coupling network layer is opposite to the coupling direction of the neurons in the second unidirectional coupling network layer. In addition, in the eight-element neural network, every two neurons are taken as a group between the first unidirectional coupling network layer and the second unidirectional coupling network layer, one group of two neurons come from different unidirectional coupling network layers, the two neurons are connected in a bidirectional way, and bidirectional coupling is formed between the two unidirectional coupling network layers, so that the equivalence of the corresponding neurons of the hip joint and the corresponding neurons of the knee joint is maintained. Four groups of neurons are respectively in one-to-one correspondence with four legs of the four-foot robot; and in each set of neurons, the neuron in the first unidirectional coupling network layer is used to control the hip joint and the neuron in the second unidirectional coupling network layer is used to control the knee joint. Thus, 8 neurons of the network can be independently controlled, and in the embodiment of the invention, the serial numbers of each group of neurons are related between the upper unidirectional coupling network layer and the lower unidirectional coupling network layer, namely, the neuron i in the first unidirectional coupling network layer and the neuron i+4 in the second unidirectional coupling network layer form a group of bi-directionally coupled neurons. In the eight-element neural network, the neuron 1 and the neuron 5 are respectively used for controlling the hip joint and the knee joint of the left rear leg, the neuron 2 and the neuron 6 are respectively used for controlling the hip joint and the knee joint of the right rear leg, the neuron 3 and the neuron 7 are respectively used for controlling the hip joint and the knee joint of the right front leg, and the neuron 4 and the neuron 8 are respectively used for controlling the hip joint and the knee joint of the left front leg. In actual implementation, the bidirectional coupling between neurons in a group can also be obtained by adding self-coupling to each neuron in a four-neuron network, so that the whole network still maintains the D 4 symmetry.
Thus, with continued reference to FIG. 3, in the eight-element neural network described above, the coupling between neurons is inhibitory, with neuron 1 inhibiting neuron 3, neuron 3 inhibiting neuron 2, neuron 2 inhibiting neuron 4, neuron 4 inhibiting neuron 1, neuron 5 inhibiting neuron 8, neuron 8 inhibiting neuron 6, neuron 6 inhibiting neuron 7, neuron 7 inhibiting neuron 5, bi-directional inhibition between neuron 1 and neuron 5, bi-directional inhibition between neuron 2 and neuron 6, bi-directional inhibition between neuron 3 and neuron 7, and bi-directional inhibition between a group of neurons 4 and 8. There are four types of coupling in total in the eight-element neural network:
First type: coupling in the top level hip neurons (corresponding to the subsequent parameter α)
Second type: coupling in underlying knee neurons (corresponding to subsequent parameter β)
Third type: coupling from top layer to bottom layer (corresponding to subsequent parameter gamma)
Fourth type: coupling from bottom layer to top layer (delta for subsequent parameters)
To maintain D 4 global symmetry, α and β should be equal, making neurons in the same layer equivalent. Gamma and delta should also be theoretically equal to maintain the equivalence of a pair of knee-hip neurons. Thus, the global symmetry of the network is formed by α and β, and the local symmetry is formed by γ and δ. But the invention can further optimize and adjust gamma and delta to meet the requirement of motion control of the quadruped robot.
In addition, in CPG is a small set of neural networks formed by interactively coupled connections, the connections between neurons are called synapses. In order to achieve motion control of neurons, a set of neuron models described by ordinary differential equations is required to model the behavior of neurons, and a coupling matrix is used to describe the connections between neurons. In the above eight-element neural network, because different coupling conditions exist, in the ordinary differential equation corresponding to the eight-element neural network, the driving signal of each neuron is determined by three parts, which are respectively gait control parameters generated by the signal regulation and control module, the coupling effect of the neurons from the same unidirectional coupling network layer, and the coupling effect of the neurons from another unidirectional coupling network layer. Thus, different neuron couplings can be considered in the ordinary differential equation, so that eight-degree-of-freedom coordinated control of the hip joint and the knee joint of the quadruped robot is realized, and five gait rhythm signals are realized.
In the embodiment of the invention, taking any neuron i in the eight-element neural network as an example, i=1, 2, … and 8 are used for demonstrating the construction and optimization modes of the neuron model.
In the eight-element neural network, a Stein neuron model is adopted by a normal differential equation of any neuron i. The Stein neuron model belongs to the prior art, and the equation formula can be expressed as follows:
Wherein, Represent the firstThe membrane potential of the individual neurons,、、Are all state quantities to be solved in the equation, but only the state quantities areThe output of the neuron is further input into a subsequent signal post-processing module.、、Respectively is、、The first derivative of time t. In a two-layer network of eight neurons, since 8 neurons are located in two unidirectional coupled network layers, some parameters need to be scaled by superscriptAndTo distinguish between a first unidirectional coupling network layer controlling the hip joint and a second unidirectional coupling network layer controlling the knee joint, respectively.Is the rate constant that affects the frequency of the neurons,AndRepresenting parameters corresponding to neurons in the first unidirectional coupling network layer and the second unidirectional coupling network layer respectively，Are driving signals of neurons and can be equally divided intoAnd。Is an adaptive constant (being a super parameter) that determines the degree of adaptation of the neuron.AndIs the rate constant (being a super parameter) of the cumulative conversion of sodium ions.
Nodes are represented in the network as a model of neurons, and connections between nodes can be understood as the mutual coupling between neurons. In the eight-element neural network of the invention, because of four different coupling conditions among neurons, in the Stein neuron model, the coupling effect from other neurons should also be included in the driving signal of the ordinary differential equationIs a kind of medium. Therefore, the invention needs to modify and reform the ordinary differential equation of a single Stein neuron so as to construct a control equation of an eight-element network.
In the embodiment of the invention, the ordinary differential equation modifies the driving signal of any neuron i into the driving signal of any neuron i based on the Stein neuron modelWherein、、Are gait control parameters generated by the signal regulation and control module, t is time,AndThe method comprises the steps of respectively obtaining the same-layer weight super parameter and the different-layer weight super parameter, wherein X is the membrane potential sum of neurons which form inhibition on the neurons i in a unidirectional coupling network layer where the current neurons i are located, and Y is the membrane potential sum of neurons which form inhibition on the neurons i in the unidirectional coupling network layer without the current neurons i.
For ease of description and easier understanding, the foregoing superscripts have been introducedAndDistinguishing the driving signal equation of the neuron i in the upper and lower unidirectional coupling network layers, and superparameter the same-layer weight in the first unidirectional coupling network layerDifferent layer weight super parameterIs marked asAndThe same layer weight in the second unidirectional coupling network layer is subjected to super parameterDifferent layer weight super parameterIs marked asAnd. Thus, the driving signal equationCan be further expressed as:
Wherein the method comprises the steps of (Corresponding to two layers of networks are respectively) Is an amplitude parameter of the entire drive signal,(Corresponding to two layers of networks are respectively) And(Corresponding to two layers of networks are respectively) Is the amplitude and frequency of the drive signal in the gait control parameters from the signal conditioning module. Summation termAndIs a coupling effect from neurons in the same layer,AndRefers to the coupling effect from neurons in other layers, whereFor indicating whether neuron j has a unidirectional coupling effect on neuron i, if soIf otherwise. Thus, from the coupling relationship between 8 neurons in the eight-element neural network shown in FIG. 3, an 8×8 coupling matrix as shown in Table 2 can be constructedTo record the one-way coupling relationship between neurons.
Table 2 coupling matrix for eight-element neural network
Parameters in the above equations (3) and (4)AndThe gait of the quadruped robot which is finally controlled by the eight-element neural network can be adjusted, and the parameters, namely the gait control parameters, can be generated by the signal regulation and control module and transmitted into the eight-element neural network.
In addition, to ensure eight-element network architectureGlobal symmetry, when solving ordinary differential equation, the same-layer weight superparameter of ordinary differential equation corresponding to eight neurons is set to be the same, namely alpha=beta is satisfied, and the different-layer weight superparameter absolute value of ordinary differential equation corresponding to neurons in the second unidirectional coupling network layer is larger than that of ordinary differential equation corresponding to neurons in the first unidirectional coupling network layer, namely。
In an embodiment of the invention, the model parameters of the network are preferably set as shown in table 3. The coupling parameters α and β are equal and are therefore both set to-0.15, the negative sign indicating inhibitory coupling between neurons. For local symmetry, ifWill result in the generation of a pair of knee-hip neuronsSymmetry. Even if the other control parameters for the neurons are not exactly equal, they still bring about a phase lock of approximately 1/2 period, since they are located in different layers with opposite coupling loops. Thus, in the eight-element neural network of the present embodiment,AndAre set to-0.6 and-0.1, respectively. This corresponds to the biological assumption that the upper network controls the gait of the whole eight-element network, the lower network following the signals generated by the upper network. And is provided withWill break the globalSymmetry, since it results in a pair of knee-hip neurons that are not exactly equivalent, in this embodimentAndAnd not the same.
Table 3 parameter categories for eight-element networks
In the embodiment of the present invention, the ordinary differential equations corresponding to the eight neurons can be solved by any feasible numerical solution method.
Thus, the motion control process of the four-legged robot according to the present invention can be described as follows: through regulating gait control parameters in the signal regulating moduleAnd) The eight-element neural network CPG in the signal generation module generates corresponding gait signals and can be sent to the signal post-processing module for processing. In general, the gait rhythm signal of each neuron solved in the signal generation module is a time domain signal with a value ranging from 0 to 1, and the control signal actually performed on the system is a joint angle signal or a voltage signal, so the signal processing layer is required to perform mapping processing and transmitting on the rhythm signal of the signal generation layer. Thus, the time domain signal may be converted by the signal post-processing module into a joint position signal or a voltage signal that controls the corresponding joint motion. The process belongs to a mature technology in robot control, and is not described in detail.
In order to demonstrate the advantages of the present invention over the prior art, the technical effects achieved by the present invention are demonstrated below by means of specific preferred examples.
In a preferred example of the present invention, the numerical simulation of an eight-element neural network is written in a Python 3.8 environment, and the ordinary differential equation of the network is calculated by a fourth-order longgrid-base tower numerical solution method. The initial state parameters of the eight-element neural network and the control parameters corresponding to the five gaits generated by the eight-element neural network are shown in table 4 and table 5 respectively, so that five gaits can be generated: walking, jogging, running, jumping and leaping.
TABLE 4 eight-element neural network initial State
Table 5 eight-element neural network control parameters for performing five gait
Finally, the results of the numerical simulation of five gaits executed by the eight-element network are shown in fig. 4, and (a) - (e) respectively show the output of knee hip neuron CPG signals and the phase diagrams among knee hip neurons under the five gaits of the eight-element neural network, wherein the phase diagrams among the knee hip neurons can prove that the eight-element network can generate stable knee hip neuron phase locking, and (f) shows the test results after disturbance injection, so that the invention has higher stability and interference resistance. In addition, the eight-element neural network is applied to simulation of the four-legged robot, and the feasibility of the technical scheme is also proved.
In conclusion, compared with a common quaternary network, the eight-element neural network provided by the invention can realize more steps and coordinate control of hip and knee joints, so that the motion control system of the four-foot robot has better environmental adaptability and maneuverability. And different from the common model-based control and learning-based control methods of the quadruped robot, the eight-element neural network architecture provided by the invention has the advantages of simple structure and small calculation amount, and can generate different gait types by adjusting the model parameters of the signal regulation and control layer.
Finally, it should be noted that in the embodiments provided by the present application, the different modules in the motion control system of the four-foot robot based on the eight-element neural network may be essentially executed by a computer program. In addition, in the embodiments provided in the present application, the division of the modules in the system is only one logic function division, and there may be another division manner in actual implementation, for example, multiple modules may be combined or may be integrated together, and one module may also be split. Programs corresponding to these modules may be stored in a computer readable storage medium in the form of software functional modules written in various programming languages for sale or use as a stand alone product. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the modules described in the various embodiments of the present application, thereby implementing the overall system functions. It is further understood that the storage medium may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Meanwhile, the storage medium may be various media capable of storing program codes, such as a USB flash disk, a mobile hard disk, a magnetic disk or an optical disk. The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but may also be a digital signal processor (DIGITAL SIGNAL Processing, DSP), application SPECIFIC INTEGRATED Circuit (ASIC), field-Programmable gate array (Field-Programmable GATE ARRAY, FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components.
The above embodiment is only a preferred embodiment of the present invention, but it is not intended to limit the present invention. Various changes and modifications may be made by one of ordinary skill in the pertinent art without departing from the spirit and scope of the present invention. Therefore, all the technical schemes obtained by adopting the equivalent substitution or equivalent transformation are within the protection scope of the invention.
Claims (9)
1. Four-foot robot motion control system based on eight-element neural network, characterized by comprising:
The signal regulation and control module is used for generating gait control parameters;
The signal generation module is used for receiving the gait control parameters generated by the signal regulation and control module, inputting the gait control parameters into the eight-element neural network, and respectively solving ordinary differential equations of the eight neurons to obtain gait rhythm signals; the eight-element neural network consists of a first unidirectional coupling network layer and a second unidirectional coupling network layer which have quadruple rotational symmetry respectively; the first unidirectional coupling network layer is formed by sequentially connecting a first neuron, a third neuron, a second neuron and a fourth neuron end to form a unidirectional coupling annular network, the second unidirectional coupling network layer is formed by sequentially connecting a fifth neuron, an eighth neuron, a sixth neuron and a seventh neuron end to form a unidirectional coupling annular network, and the coupling direction of the neurons in the first unidirectional coupling network layer is opposite to the coupling direction of the neurons in the second unidirectional coupling network layer; two-by-two neurons are used as a group between the first unidirectional coupling network layer and the second unidirectional coupling network layer, bidirectional coupling is formed between the two unidirectional coupling network layers, and four groups of neurons respectively control four legs of the four-legged robot in a one-to-one correspondence manner; and in each group of neurons, the neurons in the first unidirectional coupling network layer are used for controlling the hip joint, and the neurons in the second unidirectional coupling network layer are used for controlling the knee joint; in the ordinary differential equation corresponding to the eight-element neural network, the driving signal of each neuron is determined by three parts, namely gait control parameters generated by a signal regulation and control module, the coupling effect of the neurons from the same unidirectional coupling network layer and the coupling effect of the neurons from another unidirectional coupling network layer; in the eight-neuron neural network, a first neuron and a fifth neuron are in a group and are respectively used for controlling the hip joint and the knee joint of the left rear leg, a second neuron and a sixth neuron are in a group and are respectively used for controlling the hip joint and the knee joint of the right rear leg, a third neuron and a seventh neuron are in a group and are respectively used for controlling the hip joint and the knee joint of the right front leg, and a fourth neuron and an eighth neuron are in a group and are respectively used for controlling the hip joint and the knee joint of the left front leg;
the signal post-processing module is used for receiving the gait rhythm signal obtained by solving in the signal generating module and converting the gait rhythm signal into a displacement signal which correspondingly controls eight joints on four legs of the four-legged robot to actuate.
2. The eight-element neural network based motion control system of a quadruped robot of claim 1, wherein the gait pattern controllable by the quadruped robot includes walking, jogging, running, jumping and leaping.
3. The eight-neuron neural network based motion control system of a four-foot robot of claim 1, wherein the eight-neuron neural network uses inhibitory coupling for all couplings between neurons.
4. The eight-neuron-network-based quadruped robot motion control system according to claim 3, wherein the ordinary differential equation of any one neuron i in the eight-neuron-network adopts a Stein neuron model, and the driving signal of any one neuron i is thatWherein/>、/>、/>Are gait control parameters generated by the signal regulation and control module,/>For time,/>And/>The method comprises the steps of respectively obtaining the same-layer weight super parameter and the different-layer weight super parameter, wherein X is the membrane potential sum of neurons which form inhibition on the neurons i in a unidirectional coupling network layer where the current neurons i are located, and Y is the membrane potential sum of neurons which form inhibition on the neurons i in the unidirectional coupling network layer without the current neurons i.
5. The eight-element neural network-based motion control system of a four-foot robot of claim 4, wherein when the ordinary differential equation is solved, the same-layer weight super-parameters of the ordinary differential equations corresponding to the eight neurons are all set to be the same, and the different-layer weight super-parameters of the ordinary differential equations corresponding to the neurons in the second unidirectional coupling network layer have absolute values larger than those of the ordinary differential equations corresponding to the neurons in the first unidirectional coupling network layer.
6. The eight-neuron neural network based motion control system of a four-foot robot of claim 5, wherein the same-layer weight hyper-parameters of the ordinary differential equations for the eight neurons are all set to-0.15.
7. The eight-element neural network based motion control system of a four-foot robot of claim 5, wherein the outlier weight super parameters of the ordinary differential equations corresponding to the neurons in the first unidirectional coupling network layer are all set to-0.1, and the outlier weight super parameters of the ordinary differential equations corresponding to the neurons in the second unidirectional coupling network layer are all set to-0.6.
8. The eight-element neural network-based motion control system of a four-foot robot of claim 1, wherein the ordinary differential equation is numerically solved by a fourth-order longgrid tower method.
9. The eight-element neural network-based motion control system of the quadruped robot according to claim 1, wherein the gait rhythm signal of each neuron solved by the signal generation module is a time domain signal with a numerical range between 0 and 1, and the signal post-processing module converts the time domain signal into a joint position signal or a voltage signal for controlling the corresponding joint to actuate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410320353.3A CN117930663B (en) | 2024-03-20 | 2024-03-20 | Motion control system of four-foot robot based on eight-element neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410320353.3A CN117930663B (en) | 2024-03-20 | 2024-03-20 | Motion control system of four-foot robot based on eight-element neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117930663A CN117930663A (en) | 2024-04-26 |
| CN117930663B true CN117930663B (en) | 2024-06-04 |
Family
ID=90752389
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410320353.3A Active CN117930663B (en) | 2024-03-20 | 2024-03-20 | Motion control system of four-foot robot based on eight-element neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117930663B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1490134A (en) * | 2003-09-19 | 2004-04-21 | 清华大学 | Method and device for controlling robot simulating animal foot movement |
| CN102147592A (en) * | 2010-02-10 | 2011-08-10 | 中国科学院自动化研究所 | Fuzzy controller for controlling motion of four-footed robot |
| CN103092197A (en) * | 2011-10-28 | 2013-05-08 | 同济大学 | Four-foot robot working space track generating method based on certified program generator (CPG) mechanism |
| CN103203746A (en) * | 2012-09-29 | 2013-07-17 | 同济大学 | Method for constructing central pattern generator (CPG) control network topology structure of biped robot |
| CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
| JP2017046977A (en) * | 2015-09-02 | 2017-03-09 | 国立大学法人信州大学 | Synchronization control method of robotic wear using neural oscillator, computer program for synchronization control, and robotic wear |
| JP2019005250A (en) * | 2017-06-26 | 2019-01-17 | 国立大学法人信州大学 | Robotic wear and method for controlling the same |
| CN109324510A (en) * | 2018-09-19 | 2019-02-12 | 北京理工大学 | A kind of building of quadruped robot CPG control network, parameter tuning method |
| CN109352644A (en) * | 2018-09-14 | 2019-02-19 | 天津大学 | Manipulator motion rhythm detection method based on CPG neuroid |
| EP3854881A1 (en) * | 2020-01-22 | 2021-07-28 | Beiersdorf Ag | Classifying subjects based on their biological response to uv irradiation |
| CN116136692A (en) * | 2021-11-17 | 2023-05-19 | 南京理工大学 | Gait control method of bipedal CPG robot inspired by basal ganglia |
| CN117452817A (en) * | 2023-10-30 | 2024-01-26 | 浙江大学湖州研究院 | Gait transition and topography self-adaptive control system for adhesion multi-legged robot |
-
2024
- 2024-03-20 CN CN202410320353.3A patent/CN117930663B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1490134A (en) * | 2003-09-19 | 2004-04-21 | 清华大学 | Method and device for controlling robot simulating animal foot movement |
| CN102147592A (en) * | 2010-02-10 | 2011-08-10 | 中国科学院自动化研究所 | Fuzzy controller for controlling motion of four-footed robot |
| CN103092197A (en) * | 2011-10-28 | 2013-05-08 | 同济大学 | Four-foot robot working space track generating method based on certified program generator (CPG) mechanism |
| CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
| CN103203746A (en) * | 2012-09-29 | 2013-07-17 | 同济大学 | Method for constructing central pattern generator (CPG) control network topology structure of biped robot |
| JP2017046977A (en) * | 2015-09-02 | 2017-03-09 | 国立大学法人信州大学 | Synchronization control method of robotic wear using neural oscillator, computer program for synchronization control, and robotic wear |
| JP2019005250A (en) * | 2017-06-26 | 2019-01-17 | 国立大学法人信州大学 | Robotic wear and method for controlling the same |
| CN109352644A (en) * | 2018-09-14 | 2019-02-19 | 天津大学 | Manipulator motion rhythm detection method based on CPG neuroid |
| CN109324510A (en) * | 2018-09-19 | 2019-02-12 | 北京理工大学 | A kind of building of quadruped robot CPG control network, parameter tuning method |
| EP3854881A1 (en) * | 2020-01-22 | 2021-07-28 | Beiersdorf Ag | Classifying subjects based on their biological response to uv irradiation |
| CN116136692A (en) * | 2021-11-17 | 2023-05-19 | 南京理工大学 | Gait control method of bipedal CPG robot inspired by basal ganglia |
| CN117452817A (en) * | 2023-10-30 | 2024-01-26 | 浙江大学湖州研究院 | Gait transition and topography self-adaptive control system for adhesion multi-legged robot |
Non-Patent Citations (2)
| Title |
|---|
| 刘伟龙.基于深度强化学习的四足机器人运动规划方法研究.2024,(第03期),全文. * |
| 四足机器人对角小跑步态控制策略分析;雷静桃;贾光辉;;上海大学学报(自然科学版);20171231(第06期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117930663A (en) | 2024-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Thor et al. | Generic neural locomotion control framework for legged robots | |
| CN112060082B (en) | Online stable control humanoid robot based on bionic reinforcement learning type cerebellum model | |
| Wu et al. | CPG-based control of serpentine locomotion of a snake-like robot | |
| Li et al. | Gait planning and stability control of a quadruped robot | |
| Paine et al. | Motor primitive and sequence self-organization in a hierarchical recurrent neural network | |
| CN112051735A (en) | Humanoid robot walking control method | |
| Barron-Zambrano et al. | Perception-driven adaptive CPG-based locomotion for hexapod robots | |
| Thor et al. | Versatile modular neural locomotion control with fast learning | |
| Schilling et al. | Grounding an internal body model of a hexapod walker control of curve walking in a biologically inspired robot | |
| Jakimovski | Biologically inspired approaches for locomotion, anomaly detection and reconfiguration for walking robots | |
| Téllez et al. | Evolving the walking behaviour of a 12 dof quadruped using a distributed neural architecture | |
| Wang et al. | Efficient learning of robust quadruped bounding using pretrained neural networks | |
| Kobayashi et al. | Reinforcement learning for quadrupedal locomotion with design of continual–hierarchical curriculum | |
| CN117930663B (en) | Motion control system of four-foot robot based on eight-element neural network | |
| Pinto et al. | Modelling gait transition in two-legged animals | |
| Deng et al. | Cpg-inspired gait generation and transition control for six wheel-legged robot | |
| Pierris et al. | Learning robot control using a hierarchical som-based encoding | |
| Schilling et al. | Hierarchical MMC networks as a manipulable body model | |
| Heikkonen et al. | Self-organization and autonomous robots | |
| Li et al. | Agile and versatile bipedal robot tracking control through reinforcement learning | |
| Herrmann | Dynamical systems for predictive control of autonomous robots | |
| Liu et al. | Locomotion gait control of snake robots based on a novel unified CPG network model composed of Hopf oscillators | |
| Youcef et al. | Control of the trajectory of a hexapod robot based on distributed Q-learning | |
| Pérez-López et al. | Evolutionary Deployment of Central Pattern Generators for Legged Robots Using Nengo | |
| Yang et al. | Learning Quadruped Locomotion using Bio-Inspired Neural Networks with Intrinsic Rhythmicity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |