Method for controlling speed and direction of a rat robot based on mesencephalic-medial entorhinal cortex loop

Abstract

The application discloses a rat robot speed and direction control method based on a mesencephalon-mesolimbic olfactory cortex loop. The application induces a rat robot to complete motion starting in a stationary state through adjustable micro-electric stimulation on a mesencephalon MLR motion area and a flocculus nucleus group, collects nerve signals of "speed cells" and "direction cells" in a mesolimbic olfactory cortex MEC nucleus area, adjusts a rat robot stimulation parameter instruction set, and realizes rat motion grading control. The CnF and PPN nucleus groups of the mesencephalon motion area are taken as speed regulation instruction input points, and the flocculus is taken as a direction regulation instruction input point, so as to establish a speed and direction grading regulation instruction model, and develop a new type of rat robot which can control motion speed and motion direction in real time. The application can be used for establishing a rat robot motion state grading regulation instruction set and exploring a rat motion nerve regulation mechanism.

Description

Technical Field

[0001] The present invention belongs to the field of animal robots and relates to a behavior control method based on the motor circuit and evaluation system of the animal nervous system, and more specifically to a new rat robot based on the midbrain-medial entorhinal cortex nucleus. Background Art

[0002] Animal robots are a new type of hybrid robot that uses experimental animals as carriers and combines them with micro-electric control modules to implement instructions. The implementation method is to use the environmental perception, decision-making and movement abilities of animals to design an implantable brain-computer interface system, implant electrodes in the animal's brain, and control the animal's behavior by stimulating specific brain regions. The animals currently used include rats, fish, pigeons, beetles, bees, etc. Among them, rats are chosen as the main research objects of animal robots for the following advantages: (1) Rats are mammals and have a high genetic homology with humans. They have strong learning and memory abilities and are easy to control. After training, they can complete map recognition tasks and information collection tasks in complex environments. (2) Rats are smaller than other mammals and have good concealment, which can help people complete exploration operations. (3) Rats have strong reproductive and adaptability and can quickly adapt to harsh environments and complete exploration tasks. Based on the above advantages, animal robots can well perceive changes in the surrounding environment and complete environmental data collection. They have great research potential in the fields of social search and rescue, scientific research, etc. At the same time, animal robots are hybrid robots jointly developed based on multiple disciplines such as neuroscience, artificial intelligence, microelectronics technology, information technology, and mechanical engineering, and they have a good promoting effect on the coordinated development of multiple disciplines.

[0003] Precise control and smooth transitions of animal robots are key issues in their use. Currently, brain-computer interface technology has been used to manipulate rats' movements, such as forward and backward movements, and up and down slopes. However, traditional rat robot control methods require the experimenter to spend a significant amount of time and effort to train the rats, which is time-consuming. Furthermore, due to factors such as inappropriate site selection, imprecise stimulation parameters, and human subjectivity, the success rate of human control of rats is low, the task completion time is long, and the rat robot's movements are not smooth. The neural circuits that control rat motor behavior remain to be further explored. Therefore, further exploration of the neural circuits underlying rat motor behavior, selecting appropriate brain regions for stimulation, formulating precise stimulation parameter sets, and establishing robust command prediction models are the most pressing issues for achieving smooth control of animal robots.

[0004] Due to the lack of research on the neural system of rat motor areas, the researchers' method of controlling rat behavior through single electrical stimulation still has a large time and economic cost. In addition, with the current stimulation method, it is difficult to control the time it takes for the rat robot to complete the task, and the robot's task completion efficiency needs to be improved. Therefore, limited by the above research difficulties, a control method for multi-site electrical stimulation of the rat robot based on the rat's movement intention, with a parameterized instruction set for steering and speed classification, is urgently needed. Summary of the Invention

[0005] The purpose of the present invention is to provide a rat robot speed and direction control method based on the midbrain-medial entorhinal cortex loop. It is a multi-point coordinated electrical stimulation method based on the regulatory mechanism of the rat midbrain MLR (Midbrain Locomotion Region, MLR midbrain motor area) and SC (Superior Colliculus, SC superior colliculus), providing a basis for the hierarchical control of the speed and direction of the rat robot and the completion of difficult detection tasks.

[0006] The present invention provides a method for detecting the instantaneous speed and deflection angle of rat movement and the MEC neural signals in the brain by using MLR and SC as stimulation sites and MEC as neural signal detection point, and establishing a fine control instruction set based on "start, stop, accelerate, decelerate, turn left, and turn right".

[0007] The control method is implemented by the following steps:

[0008] S1: Based on the neural mechanism of the midbrain MLR motor nucleus regulating rat movement speed, the MLR was used as the speed control instruction target. Low-frequency micro-electrical stimulation was performed in the CnF (Cuneiform Nucleus, CnF cuneate nucleus) and PPN (Pedunculopontine nucleus, PPN ped ...

[0009] S2: Based on the superior colliculus' (SC) integration mechanism for retinal direction selection, the SC was used as the direction control command. Low-frequency micro-electrical stimulation was performed in the SC region, with bilateral SC as the target for "left deviation" and "right deviation" commands, serving as the locus for rat robot direction control.

[0010] S3: Based on the signal projection mechanism of MLR to MEC speed cells and the neural mechanism of medial entorhinal cortex MEC cells in the perception of rat movement speed and direction, the medial entorhinal cortex MEC was selected as the signal collection site to record the neural signals of the "speed" cell population and "head direction" cell population in the medial entorhinal cortex MEC. During micro-electrical stimulation of MLR, the neural signals of the speed cell population and the head direction cell population were decoded in real time based on the sequence variational autoencoder to predict the rat's movement intention;

[0011] S4: Using a multi-site collaborative stimulation-recording system with synchronized stimulation-recording functions, based on the projection mechanism of the midbrain-medial entorhinal cortex loop and the prediction function of MEC speed cells and head direction, a closed-loop control model for animal robots is established, which includes midbrain stimulation-signal decoding-behavior acquisition-midbrain restimulation, to improve the efficiency of regulation in different tasks.

[0012] The step S1 specifically includes:

[0013] S11: Based on the activation of the nucleus cuneata (CnF) in the midbrain motor area, rats were induced to produce high-frequency synchronous gait behavior. CnF neurons were selectively stimulated to implement the control instructions of "start" and "accelerate";

[0014] S12: Based on the PPN site-induced exploratory behavior and alternating gait behavior, PPN neurons in the peduncle bridge area were selected as the stimulation sites for the "slow down" and "stop" control commands.

[0015] S13: Based on the effects of the four stimulation parameters on the movement speed of rats, complete the speed classification (0-20 cm / s, 20-40 cm / s, 40-60 cm / s, 60-80 cm / s, 80-100 cm / s, and above 100 cm / s) regulation.

[0016] The step S2 specifically includes:

[0017] S21: The neural mechanism of SC integration of retinal direction modulation information. The left epithalamus was selected as the stimulation site for the "left bias" control command.

[0018] S22: Select the right epithalamus area as the stimulation site for the "right-biased" control instruction.

[0019] S23: Based on the effects of four stimulation parameters on the deflection angle of rats, the angle deflection was regulated by grades (0-30°, 30-60°, 60-90°, 90-120°, 120-150°, 150-180°).

[0020] The step S3 specifically includes:

[0021] S31: MEC signal acquisition and sorting, collecting electrical signals from speed cells and head direction cells (HD) in layers II and III of the medial entorhinal cortex. The electrodes for collecting neural signals in the medial entorhinal cortex use tungsten wire with high conductivity as the raw material and are covered with an insulating layer. By collecting the discharge signals (fire rate) and oscillation frequency of neurons in the entorhinal cortex, neurons that are clearly related to movement speed and direction are sorted out, and the discharge pattern shows a clear correlation with the rat's upcoming movement state;

[0022] S32: Further sorting neural signal features related to speed and direction levels, including positive speed control signals, negative speed control signals, left-bias control signals, and right-bias control signals. At the same time, a motion parameter prediction model based on multimodal sequence feature fusion is developed: the model is divided into a dual-stream encoder module, a feature fusion module, and a feature decoder. Based on the synchronously collected neural signal features of the MEC brain area and the rat's movement behavior characteristics such as movement speed and direction, the rat's movement state changes are predicted;

[0023] S33: Neural signal processing and command regulation. During the navigation process of the rat robot, based on the prediction and control method of the rat's next movement state (including direction and speed), the control instruction set is referred to to generate the stimulation instructions for the next MLR and SC. The behavioral data of the rat robot is collected by a high-speed camera and transmitted to the host computer. The rat's motor nerve signal is transmitted back to the host computer through the signal acquisition module of the backpack. The stimulation instructions exist in the host computer software. The stimulation parameters include: pulse period 0.00~100.00ms, pulse width 0.00~10.00ms, number of pulses 0~100, voltage amplitude 0.00~10.00V).

[0024] Rat brain positioning in steps S1, S2, and S3. Determine the implantation coordinates of the PPN, CnF, SC, and MEC sites with reference to the rat brain atlas. Select the PPN (AP: -7.8 mm, ML: 2 mm, DV: -7.0 mm), CnF (AP: -8.4 mm, ML: 2 mm, DV: -6.2 mm), SC (AP: -6.4 mm, ML: 2.2 mm, DV: -4.4 mm), and MEC (AP: 0.2-0.7 mm anterior to the sinus margin, ML: 4.5-4.8 mm, DV: -1.5-1.8 mm relative to the brain surface) as atlas sites. Measure the distance from the anterior to posterior bregma of the rat using a rat brain locator, and convert the atlas sites to the implantation sites for the current individual.

[0025] The step S4 specifically includes:

[0026] S41: Rat behavioral testing setup. The experimental platform includes a modified large open field, a high-speed camera, a stimulus-recording backpack, a treadmill, an eight-arm maze, and other experimental equipment. The high-speed camera is used to capture the rat's behavioral responses, including data such as movement speed, direction, deflection angle, and rat posture.

[0027] S42: Rat behavioral assessment. The S42 step includes:

[0028] A1: The treadmill is used to assess the rat's exercise capacity and speed. The treadmill has 8 channels, and the track can be set to different uniform speeds. A high-speed camera is installed on the top to record the rat's movement speed. The rats are pre-adapted to the treadmill track speed for 3 days before receiving stimulation instructions. When the rat moves on the track in synchronization with the treadmill, the rat's movement speed is considered constant at this time. Then, the stimulation backpack is used to input instructions to complete the speed control.

[0029] A2: The eight-arm maze is used to test whether the rat can complete a specified angle turning task during deflection control. The modified eight-arm maze provides specific task scenarios, including a maze with multiple angles of turning, a straight runway, and uphill and downhill slopes, to test the rat robot's ability to complete the specified task.

[0030] S43: Behavioral control of electrical stimulation in rats.

[0031] A lightweight stimulation-recording backpack was attached to the rat's back, consisting of a stimulation system (for inputting preset pulse stimulation commands) and a recording system (for collecting MEC neural signals). The backpack simultaneously stimulated the rat's MLR region, collected neural signals from the MEC, and transmitted them to a remote analysis system, which generated a hierarchical control instruction set based on the current movement speed and MEC neural signals.

[0032] The Stimulus-Recording Backpack includes the following modules:

[0033] Bluetooth transmission module: This module transmits pre-set electrical stimulation pulse commands to the stimulation module and collects MEC neural signals to the host computer. Neural signals primarily include raw data on neuronal discharge pulses associated with "speed" and "direction" at the MEC site.

[0034] Stimulation module: used to input electrical stimulation pulses into the electrodes. Stimulation pulse instructions include: voltage amplitude (0-10V), pulse interval (0-100ms), pulse width (0-100ms), and number of pulses (1-1000).

[0035] The stimulation-recording backpack is powered by a 12.5g, 3.6V, 200mAh lithium battery, which provides over 10 hours of continuous use. The backpack's stimulation module connects to external electrodes on the rat's brain via wires. By inputting different electrical stimulation pulses, the backpack stimulates specific brain sites, controlling the rat to complete commands. The backpack weighs no more than 40g.

[0036] A lightweight backpack is fixed on the rat's back to ensure that it can move freely. The input line at the front of the backpack is connected to the external electrodes in sequence to ensure that the action site of each instruction is accurate, including the six instructions of "start", "accelerate", "decelerate", "stop", "left deviation" and "right deviation". The instruction parameters (voltage amplitude, pulse interval, pulse width, and number of pulses) are set on the PC side, and the instruction input is controlled by the mobile controller to trigger the stimulation backpack to send a pulse sequence to the electrodes on the rat's head, activating neurons in specific nuclei in the brain, thereby controlling the rat to produce the desired behavior. A high-speed camera is used to collect the rat's behavioral data, including movement speed, head direction, and body posture points. The behavioral characteristics of the rats corresponding to different stimulation parameters are analyzed, and the corresponding relationship between electrical stimulation parameters (voltage amplitude, pulse interval, pulse width, and number of pulses) and movement behavior parameters (movement speed, deflection angle, body posture points, etc.) is established.

[0037] Furthermore, the rat robot was programmed with the specific command sequence required to complete the task, including "start," "accelerate," "turn left," "turn right," "decelerate," and "stop." A motion parameter prediction model, fused with multimodal sequence features, combined the rat's real-time motion data, MEC neural signal data, the previous motion control command, and preset motion control commands to predict the next stimulation command. The predicted stimulation command was input into the stimulation-recording backpack. Finally, based on the difference between the current behavioral response and the expected behavioral response, the parameters of the stimulation variables were optimized, gradually refined, and a refined and effective control command set based on neural signals was obtained.

[0038] In the present invention, multiple tests have shown that electrode implantation does not affect the basic motor ability of the animal itself.

[0039] The present invention has the following beneficial effects: 1. The present invention accurately locates the PPN, CnF, SC, and MEC nuclei in the rat brain, studies the neural circuit of the midbrain-entorhinal cortex, determines the impact of the neural circuit on rat behavior, and lays the foundation for electrical stimulation to control rat behavioral responses. 2. The present invention develops a multi-site implantable front-end stimulation-recording system, including multi-site implantable electrodes, an electrical stimulation backpack that can simultaneously achieve stimulation and signal acquisition, and realize the input of precise electrical stimulation sequences and the collection of electrical signals. 3. The present invention establishes a motion parameter prediction model with multi-modal sequence feature fusion, and based on the neural mechanism of the rat's entorhinal cortex to control movement speed, while giving electrical stimulation to the midbrain MLR and SC areas, collects rat movement characteristics and rat neural signals, and based on the previous control instructions, realizes the prediction of the next instruction, laying the foundation for closed-loop stimulation and regulation of rat robot movement. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Schematic diagram of the process of the present invention.

[0041] Figure 2 Diagram of PPN, CnF, SC, and MEC electrode implantation.

[0042] Figure 3 Motor control and behavioral acquisition system for rats

[0043] Figure 4 For rat neural signal and movement state system

[0044] The numbers in the figure represent the following:

[0045] 1- Stimulating electrode; 2- Recording electrode; 3- Central nervous system of rat movement control; 4- Wireless movement control backpack for rat; 5- Rat behavior control host computer system; 6- Rat behavior prediction model. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the embodiments of the present invention and the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] Example 1

[0048] like Figure 1 As shown in Figure 1, a rat robot speed and direction control method based on the midbrain-medial entorhinal cortex circuit is described. The method mainly includes the following four steps:

[0049] S1: Based on the neural mechanism of the midbrain MLR motor nucleus regulating rat movement speed, select the micro-electrical stimulation implantation sites for "start", "acceleration", "deceleration" and "stop";

[0050] S2: Based on the integration mechanism of retinal direction selection information by the superior colliculus, the SC (Superior Colliculus) serves as the "left deviation" and "right deviation" control instructions;

[0051] S3: Based on the neural mechanism of the medial entorhinal cortex MEC nucleus in the perception of rat movement speed and direction, the medial entorhinal cortex MEC was selected as the signal acquisition site to decode the neural signals of the speed cell group and the head direction cell group in real time to predict the rat's movement intention;

[0052] S4: Using a multi-site collaborative stimulation-recording system with synchronized stimulation-recording functions, based on the projection mechanism of the midbrain-medial entorhinal cortex loop and the prediction function of MEC speed cells and head direction, a closed-loop control model for animal robots is established, which includes midbrain stimulation-signal decoding-behavior acquisition-midbrain restimulation, to improve the efficiency of regulation in different tasks.

[0053] See also Figure 2 The steps S1, S2, and S3 of selecting rat motor nuclei and implanting electrodes include:

[0054] Based on the rat motor control nervous system 3, and compared with the existing rat Rat Brain Atlas brain positioning map, a rat brain locator was used to measure the distance from the rat's anterior fontanelle to the posterior fontanelle, and the map site was converted into the current individual's implantation site. Before implanting the electrode, the spatial position of several sites was determined, and the stimulating electrode 1 and the signal recording electrode 2 were loaded on the same integrator. Using a brain stereotaxic instrument, the rat's head was fixed on the locator to ensure that the mouth and nose could not move and the brain would not shake with the rat's body. The locator was used to locate and drill holes on the surface of the rat's skull to complete the electrode implantation process. The stimulating electrode 1 was implanted at the MLR and SC sites, and the signal acquisition electrode was implanted at the MEC site. After the implantation was completed, it was sealed with dental cement, and the rat rested for more than one week after the operation.

[0055] Furthermore, the rats were subjected to head electrode implantation surgery. According to the pre-measured sites, the electrodes were stereotactically implanted on the surface of the rat skull. The electrodes used double-stranded nickel-chromium alloy electrodes with an outer insulating layer. The length of the electrodes was determined according to the depth of the implantation site. At the same time, a recording electrode was implanted in the medial entorhinal cortex MEC. Recording electrode 2 used a tungsten wire with a diameter of 35 microns. After the implantation was completed, the implantation window was sealed with dental cement, and the rats that had recovered for one week after the operation were used for experimental preparation. After the behavioral experiment was completed, the complete brain tissue of the rat was removed by perfusion, fixation, etc., and paraffin sectioning was performed. Finally, the coordinate sites of each electrode were determined by Prussian blue staining, and then compared with the standard rat brain positioning map to adjust the electrode implantation site, depth, and angle to ensure the accuracy of the electrode implantation site. It was also determined whether the electrode position deviated during the experiment to ensure that the experimental data and presentation effects were true and effective.

[0056] The step S1 specifically includes:

[0057] S11: Based on the activation of the nucleus cuneata (CnF) in the midbrain motor area, rats were induced to produce high-frequency synchronous gait behavior. CnF neurons were selectively stimulated to implement the control instructions of "start" and "accelerate";

[0058] S12: Based on the PPN site-induced exploratory behavior and alternating gait behavior, PPN neurons in the peduncle bridge area were selected as the stimulation sites for the "slow down" and "stop" control commands.

[0059] S13: Based on the effects of four stimulation parameters on rat locomotion speed, we achieved speed control (0-20 cm / s, 20-40 cm / s, 40-60 cm / s, 60-80 cm / s, 80-100 cm / s, and above 100 cm / s). Pulse period and pulse width controlled acceleration, pulse number controlled success rate, and voltage amplitude controlled response time.

[0060] The step S2 specifically includes:

[0061] S21: The neural mechanism of SC integration of retinal direction modulation information. The left epithalamus was selected as the stimulation site for the "left bias" control command.

[0062] S22: Select the right epithalamus area as the stimulation site for the "right-biased" control instruction.

[0063] S23: Based on the effects of four stimulation parameters on the rat's deflection angle, the angle of deflection was controlled using a graded scale (0-30°, 30-60°, 60-90°, 90-120°, 120-150°, and 150-180°). The pulse period and pulse width modulated the angle, the number of pulses the deflection success rate, and the voltage amplitude the response time.

[0064] Step S3 specifically includes:

[0065] S31: Synchronous analysis of rat neural signals and behaviors: The rat's movement intention was measured by collecting neural signals from the MEC site. Speed ​​cells are related to speed regulation, and Head direction (HD) cells are related to the direction of rat movement. Electrodes were implanted into the MEC brain area, and the rats were given a week of postoperative recovery time. One week later, the rats were placed in a 1m*1m*1m open field in the laboratory and allowed to move freely. The real-time neural signal acquisition and small animal trajectory tracking system were used to collect the rat's movement trajectory, instantaneous movement speed, head direction, and various body posture nodes, and the neural signals of the MEC site in the rat's brain were recorded at the same time;

[0066] S32: Isolate the firing patterns unique to speed cells and HD cells. By synchronously collecting motor behavior representations and neural signals from these two types of cells, we develop a correlation model between these signals and motor states.

[0067] S33: Neural signal processing and command regulation. During the navigation process of the rat robot, based on the prediction and control method of the rat's next movement state (including direction and speed), the control instruction set is referred to to generate the stimulation instructions for the next MLR and SC. The behavioral data of the rat robot is collected by a high-speed camera and transmitted to the host computer. The rat's motor nerve signal is transmitted back to the host computer through the signal acquisition module of the backpack. The stimulation instructions exist in the host computer software. The stimulation parameters include: pulse period 0.00~100.00ms, pulse width 0.00~10.00ms, number of pulses 0~100, voltage amplitude 0.00~10.00V).

[0068] Furthermore, the step S32 specifically includes:

[0069] A1: First, the neural signal sequence S N and behavioral signal sequence S M The two streams are input into the dual-stream sequence encoder module and mapped to their respective latent spaces through sequence encoding, as shown in the following formula:

[0070]

[0071] where N neural and N motionThese are the neural signal processing part and the behavioral signal processing part of the dual-stream sequence encoder module. The two features are then mapped to the same feature space through the feature fusion module and fused in the channel dimension. Finally, the fused feature expected action (speed and direction) and the current stimulus instruction parameter set are used as input and regressed into the stimulus instruction parameter space through the contrast encoding module Reg, achieving stimulus instruction prediction based on neural and behavioral signals.

[0072] A2: Furthermore, during the training phase, each set of training data consists of a Json-formatted data packet consisting of a synchronously collected neuro-behavioral sequence signal group, an expected action, and a current instruction parameter group. Its label is the stimulation instruction parameter (voltage (V), period (ms), interval (ms), and number of stimuli (n)) for achieving the expected speed regulation, which is a one-dimensional vector of length 4. The model is trained by cross-entropy loss, and the optimization algorithm of the model is Adam. In addition, in the feature fusion module, in order to better fuse neural and behavioral features, the optimization algorithm of the model is Adam. We also added a Maximum Mean Discrepancy Loss (MMD) constraint term (the formula is as follows):

[0073]

[0074] Step S4 specifically includes: S41 Rat electrical stimulation experimental platform construction

[0075] The rat electrical stimulation control platform consists of a rat behavioral data acquisition system, a mobile control platform, a PC, and cameras. The behavioral data acquisition system includes an open field, a balance beam, an eight-arm maze, and cameras to collect behavioral response data on the rat robot, including basic motor skills, balance, post-operative recovery, and post-controlled movement speed, steering angle, and other characteristics. The mobile control platform includes a motion controller and a display to modulate wireless control commands issued by the PC. Parameters of the control command set include voltage amplitude, pulse interval, pulse width, and pulse number. The control command set is input by the mobile control platform and transmitted wirelessly via the PC. The signal is received by the rat electrical stimulation backpack in the form of pulsed voltages, which are input to the motor area of ​​the rat's midbrain via implanted electrodes. This generates a motor signal, inducing a behavioral response in the rat robot. Changes in movement state are captured by cameras positioned above, in front of, and to the side of the rat, providing detailed gait characteristics. The final behavioral data is received by the PC. Through detailed behavioral characterization and analysis, the parameter-graded, precise control command set is further optimized to improve control accuracy and smooth behavioral switching.

[0076] S42: Rat behavioral assessment, the S42 step includes:

[0077] A1: The treadmill is used to assess the rat's exercise capacity and speed. The treadmill has 8 channels, and the track can be set to different uniform speeds. A high-speed camera is installed on the top to record the rat's movement speed. The rats are pre-adapted to the treadmill track speed for 3 days before receiving stimulation instructions. When the rat moves on the track in synchronization with the treadmill, the rat's movement speed is considered constant at this time. Then, the stimulation backpack is used to input instructions to complete the speed control.

[0078] A2: The eight-arm maze is used to test whether the rat can complete a specified angle turning task during its deflection control process. The modified open field provides specific task scenarios, including a maze with multiple angles of turning, a straight runway, and uphill and downhill slopes, to test the rat robot's ability to complete the specified task.

[0079] S43: Behavioral Control of Electrical Stimulation in Rats

[0080] Connect the electrodes at the CnF site to the stimulation backpack interfaces corresponding to the "start" and "accelerate" commands, the electrodes at the PPN site to the stimulation backpack interfaces corresponding to the "slow down" and "stop" commands, and the electrodes at the left and right SC sites to the backpack interfaces corresponding to the "left deviation" and "right deviation" commands. Simultaneously, the recording interface of the lightweight backpack is connected to the signal recording electrodes at the MEC site to synchronously collect the rat's movement-related neural signals during stimulation. These signals are then transmitted to the host computer via a Bluetooth transmission system for real-time analysis. The analysis method is similar to step S3.

[0081] The step S43 specifically includes:

[0082] A1: After the rats recovered from surgery, they were given a light backpack (such as Figure 3 ), put the rat robot at the entrance of the maze, and first send the "start" command through the pre-set stimulation parameters, the stimulation electrode 1 outputs a pulse stimulation, gives low-intensity electrical stimulation to the CnF nucleus, and then induces the rat to start movement through the spinal motor neurons. When the rat's movement speed reaches more than 10cm / s, high-intensity electrical stimulation of the CnF nucleus is given to increase the excitability of neurons, thereby increasing the rat's movement speed, which can be as high as more than 100cm / s. During the movement process, when the rat encounters an obstacle, the movement speed is reduced to less than 40cm / s through low-intensity stimulation of the PPN site, and then the pre-set stimulation parameters are used to stimulate the SC brain areas of the left and right brains of the rat respectively, adjust the rat's movement deflection angle, and avoid obstacles. Finally, high-intensity electrical stimulation is given to the PPN site to induce the rat robot to stop moving;

[0083] A2: Different stimulation parameters were used to target different stimulation effects. By adjusting the pulse frequency and pulse width, we regulated the rat's movement speed (including 0-20 cm / s, 20-40 cm / s, 40-60 cm / s, 60-80 cm / s, 80-100 cm / s, and above 100 cm / s) and the rat's deflection angle (including 0-30°, 30-60°, 60-90°, 90-120°, 120-150°, and 150-180°). By adjusting the number of stimuli, we altered the duration of the rat's movement state. By adjusting the voltage amplitude, we adjusted the threshold and response time of the rat's movement state switching.

[0084] like Figure 4 During the rat's movement, the robot's motion state data is captured by a camera and transmitted to the rat behavior control host computer system 5. This data includes movement speed, movement direction, and 10 rat posture points, forming a rat movement dataset. Simultaneously, speed cells and head direction cells in the MEC (Mechanical Center for the Central Evolutionary Organisation) continuously discharge. Recording electrodes 2 implanted in the rat's motor control central nervous system 3 within the MEC collect neuronal firing rates and theta oscillation frequencies. This neuronal firing information is then transmitted to the rat behavior control host computer system 5 via the rat wireless motion control backpack 4, forming a motor nerve signal dataset related to the rat's movement.

[0085] The motion data set and the motor nerve signal data set are used as representations of the current behavior and neural mechanism of the rat. Combined with the current stimulation parameter instruction set, based on the expected behavioral task, the motion parameter prediction model based on multimodal sequence feature fusion in step S32 is used to realize stimulation parameter prediction, output the stimulation parameter prediction result, and then transmit the predicted stimulation parameter instruction set to the stimulation module in the host computer, and then transmit it to the rat wireless motion control backpack 4 using Bluetooth, and further induce the rat to complete the expected task based on electrical stimulation.

Claims

1. A rat robot speed and direction control method based on the midbrain-medial entorhinal cortex circuit, characterized in that: This is achieved by following these steps: S1: Based on the neural mechanism of the MLR motor nucleus in the midbrain motor region controlling rat movement speed, low-frequency micro-electrical stimulation was performed in the cuneate nucleus CnF and the pedunculopontine nucleus PPN, with the CnF as the target for "start and accelerate" instructions and the PPN as the target for "decelerate and stop" instructions, as the sites for rat robot speed control; S2: Based on the superior colliculus' integration mechanism of retinal direction selection information, the superior colliculus (SC) was used as the direction control instruction. Low-frequency micro-electrical stimulation was performed in the SC area, with bilateral SC as the "left deviation" and "right deviation" instruction targets, as the site for rat robot direction control; S3: Based on the signal projection mechanism of MLR to speed cells in the medial entorhinal cortex (MEC), and the encoding function of MEC cells for movement speed and direction, the neural signals of the "speed" and "head direction" cell populations in the medial entorhinal cortex (MEC) were recorded. During micro-electrical stimulation of the MLR, the neural signals of the speed and head direction cell populations were decoded in real time using a sequence variational autoencoder to predict the rat's movement intention. S4: Using a multi-site collaborative stimulation-recording system with synchronized stimulation-recording functions, based on the projection mechanism of the midbrain-medial entorhinal cortex loop and the prediction function of MEC speed cells and head direction, a closed-loop control model for animal robots is established, which includes midbrain stimulation-signal decoding-behavior acquisition-midbrain restimulation, to improve the efficiency of regulation in different tasks.

2. The control method according to claim 1, characterized in that: The step S1 comprises: S11: Based on the activation of the nucleus cuneata (CnF) in the midbrain motor region, rats were induced to produce high-frequency synchronized gait behavior. CnF neurons were selectively stimulated to implement the control instructions of "start" and "accelerate"; S12: Based on the PPN site-induced exploratory behavior and alternating gait behavior, PPN neurons in the peduncle bridge area were selected as the stimulation sites for the "slow down" and "stop" control commands; S13: Complete speed graded control, graded as 0~20 cm / s, 20~40 cm / s, 40~60 cm / s, 60~80 cm / s, 80~100 cm / s, and above 100 cm / s.

3. The control method according to claim 1, wherein: The step S2 comprises: S21: Based on the neural mechanism of SC integrating retinal direction modulation information, the left superior thalamus area was selected as the stimulation site for the "left-biased" control command; S22: Select the right epithalamus as the stimulation site for the "right-biased" control command; S23: Angle deflection graded control, grades are 0~30°, 30~60°, 60~90°, 90~120°, 120~150°, 150~180°.

4. The control method according to claim 1, wherein: The step S3 comprises: S31: MEC signal acquisition and sorting, collecting electrical signals of speed cells and head direction cells in layers II and III of the medial entorhinal cortex; S32: Further sorting neural signal features related to speed and direction levels, and simultaneously building a motion parameter prediction model based on multimodal sequence feature fusion: The model consists of a dual-stream encoder module, a feature fusion module, and a feature decoder. Based on the synchronously collected characteristics of the MEC brain neural signal changes and the speed and direction movement behavior characteristics, the model predicts the changes in the rat's movement state. S33: Neural signal processing and command regulation. During the rat robot navigation process, based on the prediction and control method of the rat's motion state at the next moment, including direction and speed, the next MLR and SC stimulation instructions are generated with reference to the control instruction set. The stimulation parameters include: pulse period 0.00~100.00 ms, pulse width 0.00~10.00 ms, number of pulses 0~100, and voltage amplitude 0.00~10.00 V.

5. The control method according to claim 1, characterized in that: The step S4 comprises: S41: Rat behavioral scenario test setup. The experimental platform includes: a modified large open field, a high-speed camera, a stimulus-recording backpack, a treadmill, and an eight-arm maze. The high-speed camera is used to collect the rat's behavioral responses. The collected data include: movement speed, movement direction, deflection angle, and rat posture point. S42: Behavioral assessment of rats, including: A1: The treadmill is used to evaluate the exercise capacity and speed of rats. A2: Eight-arm maze for angular assessment of yaw control in rats; S43: Closed-loop control system for rat behavior A lightweight stimulation-recording backpack is installed on the rat's back, including a stimulation system and a recording system. The backpack synchronously stimulates the rat's MLR area, collects neural signals from the MEC site, and transmits them to a remote analysis system to obtain a hierarchical control instruction set based on the current movement speed and MEC neural signal regulation.

Images