A bio-neural network-based all-optical intelligent computing control method and system

Abstract

This invention provides an all-optical intelligent computing control method and system based on biological neural networks. The method involves determining the position of an obstacle relative to the vehicle and the distance between the obstacle and the vehicle using sensors installed on the vehicle; determining the stimulus position based on the obstacle's position relative to the vehicle and the stimulus intensity based on the distance between the obstacle and the vehicle; applying stimulation to pre-set biological neural samples based on the stimulus position and intensity; constructing a feedback vector based on the responses of the biological neural samples; and inputting the feedback vector into a pre-trained neural network model, which then outputs vehicle commands. This solution first utilizes the application of light stimulation to pre-set biological neural samples, causing neurons to respond by changing their brightness. This approach leverages the realistic reflection of biological neural samples to improve processing efficiency and response accuracy, and combines biological neural networks with artificial intelligence, using biological neural networks to enhance the processing efficiency of artificial intelligence.

Description

Technical Field

[0001] This invention relates to the field of brain science technology, and in particular to an all-optical intelligent computing control method and system based on biological neural networks. Background Technology

[0002] With the development of artificial intelligence, it continues to iterate and evolve, opening up new functions in various directions and bringing more and more surprises and shocks.

[0003] Artificial intelligence (AI) possesses powerful capabilities, but it also has many shortcomings. For example, the growth in computing power demand has far exceeded macroeconomic trends, power consumption is gradually increasing, and its performance in handling complex tasks remains unsatisfactory. Comparing the latest supercomputer, the Frontier supercomputer, to the human brain, the former consumes around 21 MW, while the human brain only consumes 10-20 MW—a difference of six orders of magnitude. Furthermore, the supercomputer occupies 680 square meters of space. 2 The human brain weighs only about 1.3 kg, so it can be said that biological intelligence still has irreplaceable advantages: 1. The human brain can better process small or uncertain data; 2. The human brain can perform both sequential and parallel processing (while computers can usually only perform the former); 3. The human brain performs better in decision-making on large, highly heterogeneous and incomplete datasets and other challenging forms of processing.

[0004] The human nervous system is an extremely complex organization, containing nearly 86 billion neurons, each with thousands of synapses connecting it to other neurons. Neurons transmit information through synapses, and these interconnected neurons form a neural network—the nervous system. A vast number of somatic cells with sensing and stretching capabilities are connected to the input and output ends of this network structure via nerve fibers. It is through this network structure that the central nervous system acquires intelligence. However, current technologies struggle to utilize biological neural networks to improve the processing performance of artificial intelligence. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide an all-optical intelligent computing control method and system based on biological neural networks, in order to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of the present invention provides an all-optical intelligent computing control method based on a biological neural network, the method comprising the following steps:

[0007] The position of the obstacle relative to the vehicle and the distance between the obstacle and the vehicle are determined based on sensors installed on the vehicle;

[0008] The stimulus location is determined based on the position of the obstacle relative to the vehicle, and the stimulus intensity is determined based on the distance between the obstacle and the vehicle.

[0009] Stimulation is applied to a pre-set biological neural sample based on the stimulation location and intensity, and a feedback vector is constructed based on the response of the biological neural sample.

[0010] The feedback vector is input into a pre-trained neural network model, which outputs vehicle commands.

[0011] Using the above scheme, this scheme first uses sensors installed on the vehicle to determine the position and relative distance of obstacles near the vehicle, and then applies light stimulation to pre-set biological neural samples. After receiving light stimulation, the neurons in the biological neural samples will respond accordingly by changing their own brightness. Finally, the brightness of the neurons is used to construct a feedback vector, and the final vehicle command is output through a neural network model. This scheme can improve processing efficiency and response accuracy by using biological neural samples to reflect the actual situation, and combines biological neural networks with artificial intelligence to improve the processing efficiency of artificial intelligence.

[0012] In some embodiments of the present invention, in the step of determining the stimulus position based on the position of the obstacle relative to the vehicle, if the position of the obstacle relative to the vehicle is to the left of the vehicle, the stimulus position is a first position; if the position of the obstacle relative to the vehicle is to the right of the vehicle, the stimulus position is a second position.

[0013] In some embodiments of the present invention, the method provides a projection area for the biological neural sample, the projection area being rectangular. In the step of determining the stimulus position based on the position of the obstacle relative to the vehicle, the point corresponding to the first position is a point 395 μm away from both the left and upper boundaries of the projection area; the point corresponding to the second position is a point 395 μm away from both the right and upper boundaries of the projection area.

[0014] In some embodiments of the present invention, in the step of determining the stimulus intensity based on the distance between the obstacle and the vehicle, the distance between the obstacle and the vehicle is matched with the stimulus intensity based on a preset stimulus intensity correspondence to obtain a corresponding stimulus intensity value. In the stimulus intensity correspondence, the farther the distance between the obstacle and the vehicle, the lower the corresponding stimulus intensity.

[0015] In some embodiments of the present invention, in the step of applying stimulation to a pre-set biological neural sample based on the stimulation location and stimulation intensity, the point corresponding to the first or second location is used as the center of a circle, and light stimulation is applied to a circular area.

[0016] In some embodiments of the present invention, the method further includes determining a stimulation range based on the vehicle speed, matching the vehicle speed with the stimulation range based on a preset stimulation range correspondence, and obtaining a corresponding stimulation range value, wherein the faster the vehicle speed, the larger the corresponding stimulation range.

[0017] In some embodiments of the present invention, in the step of constructing a feedback vector based on the response of a biological neural sample, the feedback intensity of the neuron in the biological neural sample is used as the value of each dimension in the feedback vector to construct the feedback vector.

[0018] In some embodiments of the present invention, in the step of constructing the feedback vector by using the feedback intensity of neurons in the biological neural sample as the value of each dimension of the feedback vector, the feedback of the neurons is optical feedback, and the feedback vector is constructed by using the light intensity of the feedback from the neurons as the value of each dimension of the feedback vector.

[0019] In some embodiments of the present invention, the projection region includes a first reaction region and a second reaction region, wherein the first reaction region is a response to a stimulus applied to a first location, and the second reaction region is a response to a stimulus applied to a second location. In the step of constructing the feedback vector by using the feedback intensity of neurons in the biological neural sample as the value of each dimension of the feedback vector, the feedback intensity of neurons in the first reaction region and the second reaction region is used as the value of each dimension of the feedback vector to construct the feedback vector.

[0020] A second aspect of the present invention also provides an all-optical intelligent computing control system based on a biological neural network. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method described above.

[0021] A third aspect of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned all-optical intelligent computing control method based on biological neural networks.

[0022] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the text, or may be learned by practice of the invention. The objects and other advantages of the invention will become apparent from the description and the accompanying drawings.

[0023] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0024] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, are not intended to limit the scope of the invention.

[0025] Figure 1 This is a schematic diagram of one implementation of the all-optical intelligent computing control method based on biological neural networks in this scheme;

[0026] Figure 2 This is a schematic diagram of the overall architecture of this solution;

[0027] Figure 3 This is a schematic diagram of the data processing architecture of this solution;

[0028] Figure 4 This is a schematic diagram of the projection area of ​​the experimental example;

[0029] Figure 5 This is a schematic diagram showing the output results and weight changes of the experimental example. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0031] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0032] like Figure 1 , 2 As shown in Figure 3, this invention proposes an all-optical intelligent computing control method based on biological neural networks. The steps of this method include:

[0033] Step S100: Determine the position of the obstacle relative to the vehicle and the distance between the obstacle and the vehicle based on the sensors installed on the vehicle;

[0034] In practice, the sensors installed on the vehicle are radars. The radars are used to identify obstacles around the vehicle and determine the distance and direction of the obstacles from the vehicle.

[0035] Step S200: Determine the stimulus location based on the position of the obstacle relative to the vehicle, and determine the stimulus intensity based on the distance between the obstacle and the vehicle;

[0036] Step S300: Apply stimulation to the preset biological neural sample based on the stimulation location and stimulation intensity, and construct a feedback vector based on the response of the biological neural sample;

[0037] In the specific implementation process, the biological neural sample was a mouse cortical neuron placed in a culture dish. Cortical tissue was obtained from fetal rats at 15.5 days of gestation in Beijing Spaford mice. To ensure tissue integrity and cell viability, the entire operation was strictly performed on ice. After carefully peeling off the fetal rat cortical tissue, ensuring that each piece was of uniform size, it was cut into small pieces and immersed in DMEM (#CM15020) from Zhongke Maichen. After gently washing twice with DMEM, an appropriate amount of deoxyribonuclease I (#D8071) from Solarbio was added to 0.125% Zhongke Maichen trypsin (#CC017), and digestion was carried out at 37°C for 20 minutes. After digestion, the tissue was further dissociated by pipetting to obtain a single-cell suspension. Then, the obtained cell suspension was collected and centrifuged at 80×g for 5 minutes to separate single cells. After centrifugation, the supernatant was carefully discarded. To improve cell viability and subsequent culture results, cells were resuspended in Neurobasal Plus medium (Gibco, #A3582901) supplemented with Gibco's B27 Plus supplement (#A3582801) and GlutaMax additive (#35050061). Simultaneously, penicillin-streptomycin bispecific antibiotics (#CC004) from Zhongke Maichen were added to the medium. After resuspending the cells, the cell suspension was evenly seeded at a density of 3-4 × 10^5 cells / ml into 24-well plates (Corning, #3524) pre-coated with Sigma-Aldrich's poly-D-lysine (PDL, #P6407). The culture plates were then incubated at 37°C in a 5% CO2 incubator. On day 3 of culture (DIV3), cells were transduced using AAV2 / 9-CAG-CheRiff-eGFP-WPRE-bGHpA and AAV2 / 9-CAG-jRCaMP1b-WPRE-bGHpA adeno-associated viruses. These two viral vectors efficiently expressed the optogenetic stimulant protein CheRiff and the calcium indicator protein jRCaMP1b. After approximately 14 days of culture, neurons differentiated and matured, and subsequent light stimulation and calcium imaging experiments were performed.

[0038] Step S400: Input the feedback vector into a pre-trained neural network model, and the neural network model outputs vehicle commands.

[0039] In the specific implementation process, the vehicle instruction is to turn left, turn right, or go straight, etc. Specifically, if the vehicle instruction is to turn left or turn right, the vehicle instruction includes the angle of turning left or turning right.

[0040] In the specific implementation process, the neural network model adopts the FORCE-learning algorithm, which updates the weight of each neuron in real time, calculates the weighted sum of the discharge intensity and the weight, and generates vehicle commands.

[0041] Using the above scheme, this scheme first uses sensors installed on the vehicle to determine the position and relative distance of obstacles near the vehicle, and then applies light stimulation to pre-set biological neural samples. After receiving light stimulation, the neurons in the biological neural samples will respond accordingly by changing their own brightness. Finally, the brightness of the neurons is used to construct a feedback vector, and the final vehicle command is output through a neural network model. This scheme can improve processing efficiency and response accuracy by using biological neural samples to reflect the actual situation, and combines biological neural networks with artificial intelligence to improve the processing efficiency of artificial intelligence.

[0042] In some embodiments of the present invention, in the step of determining the stimulus position based on the position of the obstacle relative to the vehicle, if the position of the obstacle relative to the vehicle is to the left of the vehicle, the stimulus position is a first position; if the position of the obstacle relative to the vehicle is to the right of the vehicle, the stimulus position is a second position.

[0043] In some embodiments of the present invention, the method provides a projection area for the biological neural sample, the projection area being rectangular. In the step of determining the stimulus position based on the position of the obstacle relative to the vehicle, the point corresponding to the first position is a point 395 μm away from both the left and upper boundaries of the projection area; the point corresponding to the second position is a point 395 μm away from both the right and upper boundaries of the projection area.

[0044] In some embodiments of the present invention, in the step of determining the stimulus intensity based on the distance between the obstacle and the vehicle, the distance between the obstacle and the vehicle is matched with the stimulus intensity based on a preset stimulus intensity correspondence to obtain a corresponding stimulus intensity value. In the stimulus intensity correspondence, the farther the distance between the obstacle and the vehicle, the lower the corresponding stimulus intensity.

[0045] In some embodiments of the present invention, in the step of applying stimulation to a pre-set biological neural sample based on the stimulation location and stimulation intensity, the point corresponding to the first or second location is used as the center of a circle, and light stimulation is applied to a circular area.

[0046] In the specific implementation process, in the step of applying stimulation to the pre-set biological neural sample based on the stimulation location and stimulation intensity, a digital micromirror device (DMD) is used to convert the obstacle information detected by the vehicle into optical stimulation and apply it precisely to the biological neural sample.

[0047] In some embodiments of the present invention, the method further includes determining a stimulation range based on the vehicle speed, matching the vehicle speed with the stimulation range based on a preset stimulation range correspondence, and obtaining a corresponding stimulation range value, wherein the faster the vehicle speed, the larger the corresponding stimulation range.

[0048] Using the above scheme, this scheme simulates multiple stimuli to simulate real stimulation of neurons, enabling neurons to give more realistic responses. By capturing the neuronal responses through images and determining the actual responses based on neural network models, the accuracy and efficiency of the responses are guaranteed.

[0049] In some embodiments of the present invention, in the step of constructing a feedback vector based on the response of a biological neural sample, the feedback intensity of the neuron in the biological neural sample is used as the value of each dimension in the feedback vector to construct the feedback vector.

[0050] In some embodiments of the present invention, in the step of constructing the feedback vector by using the feedback intensity of neurons in the biological neural sample as the value of each dimension of the feedback vector, the feedback of the neurons is optical feedback, and the feedback vector is constructed by using the light intensity of the feedback from the neurons as the value of each dimension of the feedback vector.

[0051] In the specific implementation process, in the step of obtaining optical feedback, this scheme uses a high-precision optical camera to capture the firing information of biological neurons in real time (0.5s per frame) and read the changes in the firing information of the selected neurons.

[0052] The high-precision optical camera is a CMOS camera. Specifically, the firing of neurons causes an influx of intracellular calcium ions, which react with a fluorescent indicator (jRCaMP1b) and cause a change in fluorescence intensity.

[0053] This scheme uses LED (LED1, center wavelength: 565nm, green light) for imaging to observe neural network activity. The photosensitive protein (CheRiff) is introduced into the neural network through genetic engineering, giving it optogenetic properties. The DMD can reflect user-controlled dynamic light patterns, enabling precise spatiotemporal control of light stimulation for specific cells or all cells within a given region.

[0054] In some embodiments of the present invention, the projection region includes a first reaction region and a second reaction region, wherein the first reaction region is a response to a stimulus applied to a first location, and the second reaction region is a response to a stimulus applied to a second location. In the step of constructing the feedback vector by using the feedback intensity of neurons in the biological neural sample as the value of each dimension of the feedback vector, the feedback intensity of neurons in the first reaction region and the second reaction region is used as the value of each dimension of the feedback vector to construct the feedback vector.

[0055] This scheme employs real biological neural networks to perform reservoir computation, extracts neuronal signals, and uses the FORCE-learning algorithm to map these signals into different outputs, exploring the essence of biological intelligent learning. This scheme maps the inputs and outputs of the biological neural network to the vehicle environment, enabling real-time control of vehicle obstacle avoidance by the biological neural network. It also holds promise for handling more complex tasks and finding wider applications in fields such as brain science and computational biology.

[0056] Experimental Example

[0057] In the experimental example, the vehicle ran on two tracks and could detect the distance to an approaching obstacle directly ahead. This distance information was transmitted to the PC controlling the optical system. Based on the vehicle's track and the distance to the obstacle, different images were set for the DMD, and the LEDs were switched on and off. A center wavelength of 470nm and a light power density of 1.0mW / mm² were used. 2 -2.0mW / mm 2 The blue light is used to input obstacle information detected by the vehicle into the stimulation perception area of ​​the neural network in real time.

[0058] During the experiment, the neuron images captured had a resolution of 640*540, and a stimulus-sensing input area was set within a 200*600 pixel range (approximately 500μm*1500μm) above the captured area.

[0059] Within the sensory region, this protocol selects highly active neurons for sensing stimulus input. A circular optical stimulus is used, covering 5-10 biological neurons with a radius of approximately 145 μm. Preliminary experiments are conducted to test the responses of sensory neurons in different locations to determine the most suitable region for subsequent experiments.

[0060] After the experimental procedure begins, such as Figure 3 and 4As shown, when the vehicle encounters an obstacle on the left track, the PC adjusts the DMD image settings so that the aperture is located in the upper left region of the neuron's observation area. The aperture is 250μm away from both the left and upper edges. This aperture position allows the neural network to accurately perceive the optical stimulus input, and the LED light is kept at an appropriate distance from the calculation output area to ensure that the LED light intensity does not affect the output results of the calculation output area. When this area is stimulated, it indicates that the vehicle has detected an obstacle on the left track. Simultaneously, the PC adjusts the LED light intensity according to the distance between the vehicle and the obstacle; the closer the distance, the greater the light intensity. When the vehicle detects an obstacle at a distance of 5 vehicle lengths, the LED light power density is set to 1.0mW / mm². 2 When the vehicle detects an obstacle at a distance of one vehicle length, the LED light power density is set to 2.0 mW / mm². 2 When the vehicle is running on the right track, the principle is the same. The DMD image is adjusted so that the LED light shines on the upper right area of ​​the neuron's observation region. Similar to when an obstacle is detected on the left, the distance of the aperture from both the right and upper edges is 250 μm, indicating that the vehicle has detected an obstacle on the right track. By shining optical stimuli on different areas, different inputs are given to the biological neural network, resulting in different output responses from the biological neural network.

[0061] During the experiment, the aperture size set by the DMD was adjusted according to the vehicle's operating speed. The vehicle's operating speed was divided into four levels, from 1 to 4, corresponding to an outer ring radius of 85μm, 105μm, 125μm, and 145μm, respectively. The faster the operating speed, the larger the aperture, resulting in a wider range of optical stimulation input for the biological neural network.

[0062] like Figure 5 As shown, during the experiment, 15-20 neurons were selected as computational units in each of the first and second reaction regions. The changes in light intensity were statistically analyzed and mapped into different control commands through FORCE-learning.

[0063] During the preliminary experiments, the response of the biological neural network to stimuli of different light intensities was tested simultaneously. By controlling the value of the stimulus light intensity, the biological neural network can be made sensitive to the stimulus without being affected by excessive light intensity. The appropriate range for light power density is concentrated in the range of 1.5 mW / mm². 2 -2.0mW / mm 2 .

[0064] In the experimental procedure, the intelligent computing system captures images of neuronal optical imaging in real time at a rate of 0.5 seconds per frame. It reads the firing intensity of selected neurons in two computational execution regions, multiplies it by the weight of each neuron, and sums the results to calculate the output of each region. When the vehicle detects an obstacle on one side, it will exhibit different movement outcomes based on the output of the two regions: 1. If the result of the computational execution region corresponding to the obstacle detection region exceeds a preset threshold, and the result of the computational execution region on the other side does not exceed the preset threshold, the biological neural network outputs a "turn" command, and the vehicle turns; 2. If the results of both computational execution regions exceed the preset threshold, the biological neural network cannot determine whether to turn left or right, and the vehicle remains stationary; 3. If the results of neither computational execution region exceed the preset threshold, the biological neural network does not issue a movement command, and the vehicle also remains stationary.

[0065] This scheme utilizes optogenetics to precisely detect the emitted signals of a neural network, significantly improving spatial resolution. It also defines two output units within the neural network, demonstrating the ability of biological neural networks to perform multiple computations simultaneously. For example, two computation execution regions are set up, both employing FORCE-learning to extract continuous signals from a reservoir to control the left and right steering of a virtual car to avoid obstacles. After FORCE-learning training, the system's success rate exceeds 95%.

[0066] This scheme realizes an intelligent closed-loop control system: it performs image analysis on the experimentally cultured real biological neural network, identifies individual neurons, defines the perception area and stimulation area, extracts the light stimulation and firing intensity of neurons, and establishes a mapping relationship between the biological neural network and the virtual car experimental environment to realize real-time control of the game; it also explores the learning mechanism of the biological neural network in depth: it integrates the real-time operation status of the virtual car and its surrounding environment with the dynamic input and output of the biological neural network, revealing and simulating the process by which the biological nervous system processes complex perceptual tasks and learns from them.

[0067] This invention also provides an all-optical intelligent computing control system based on a biological neural network. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method described above.

[0068] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned all-optical intelligent computing control method based on biological neural networks. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.

[0069] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0070] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0071] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0072] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fully optical intelligent computing control method based on biological neural networks, characterized in that, The steps of this method include: The position of the obstacle relative to the vehicle and the distance between the obstacle and the vehicle are determined based on sensors installed on the vehicle; The stimulus location is determined based on the position of the obstacle relative to the vehicle, and the stimulus intensity is determined based on the distance between the obstacle and the vehicle. Stimulation is applied to a pre-set biological neural sample based on the stimulation location and intensity, wherein the biological neural sample is a mouse cortical neuron set in a culture dish; A feedback vector is constructed based on the response of biological neural samples. The feedback intensity of neurons in the biological neural samples is used as the value of each dimension of the feedback vector to construct the feedback vector. The feedback of neurons is optical feedback. The light intensity of the feedback from neurons is used as the value of each dimension of the feedback vector to construct the feedback vector. The feedback vector is input into a pre-trained neural network model, which outputs vehicle commands.

2. The all-optical intelligent computing control method based on biological neural networks according to claim 1, characterized in that, In the step of determining the stimulus position based on the position of the obstacle relative to the vehicle, if the obstacle is on the left side of the vehicle, the stimulus position is the first position; if the obstacle is on the right side of the vehicle, the stimulus position is the second position.

3. The all-optical intelligent computing control method based on biological neural networks according to claim 2, characterized in that, The method sets a projection area for the biological neural sample. The projection area is rectangular. In the step of determining the stimulation position based on the position of the obstacle relative to the vehicle, the point corresponding to the first position is a point that is 395 μm away from both the left and upper boundaries of the projection area; the point corresponding to the second position is a point that is 395 μm away from both the right and upper boundaries of the projection area.

4. The all-optical intelligent computing control method based on biological neural networks according to claim 1, characterized in that, In the step of determining the stimulus intensity based on the distance between the obstacle and the vehicle, the distance between the obstacle and the vehicle is matched with the stimulus intensity based on a preset stimulus intensity correspondence to obtain the corresponding stimulus intensity value. In the stimulus intensity correspondence, the farther the distance between the obstacle and the vehicle, the lower the corresponding stimulus intensity.

5. The all-optical intelligent computing control method based on biological neural networks according to claim 3, characterized in that, In the step of applying stimulation to a pre-set biological neural sample based on the stimulation location and stimulation intensity, the point corresponding to the first or second location is used as the center of a circle, and light stimulation is applied to a circular area.

6. The all-optical intelligent computing control method based on biological neural networks according to claim 1, characterized in that, The method further includes determining the stimulation range based on the vehicle's speed, matching the vehicle's speed with the stimulation range based on a preset stimulation range correspondence, and obtaining the corresponding stimulation range value, wherein the faster the vehicle's speed, the larger the corresponding stimulation range.

7. The all-optical intelligent computing control method based on biological neural networks according to claim 3, characterized in that, The projection region includes a first reaction region and a second reaction region. The first reaction region is the response to a stimulus applied to a first location, and the second reaction region is the response to a stimulus applied to a second location. In the step of constructing the feedback vector by using the feedback intensity of neurons in the biological neural sample as the value of each dimension of the feedback vector, the feedback intensity of neurons in the first and second reaction regions is used as the value of each dimension of the feedback vector to construct the feedback vector.

8. A fully optical intelligent computing control system based on biological neural networks, characterized in that, The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method as described in any one of claims 1 to 7.

Images