Mobile robots can replace people into dangerous environments such as fire and earthquake sites for terrain exploration and casualty search, but most robots are difficult to adapt to obstacles, right-angle walls, wall transitions and other complex terrain at the same time, and their control systems are relatively complex, requiring external energy input. Therefore, a deformable mobile robot with multiple functions, relatively simple control and no tethering was designed. Firstly, taking the 2-bar 4-cable tensioning integral structure as the basic unit, the body of the tensioning integral structure which could realize bending deformation was designed. Secondly, based on the analysis of adsorption force and structural parameters, the negative pressure adsorption device was designed and combined with the deformable body to form the overall structure of the robot. Then, the kinematic analysis of the robot was carried out, and the mapping relationship between the robot pose and the motor angle was obtained. Based on this, the robot's gaits of wall surface transition, traversing the narrow space from the wall surface and flipping up steps were planned. Finally, the robot prototype was developed, and the robot movement experiments were carried out according to different terrain, and the rationality of the robot gait planning was verified. The research results provide a certain reference value for the design and manufacture of multi-functional mobile robots.

Because of the asymmetric structure of axial piston pump, its output pressure and output flow have pulsating characteristics, which affects the output stability and reliability of hydraulic system. Therefore, an optimization design method of low pulsation structure of swashplate axial piston pump based on multi-objective genetic algorithm is proposed. Firstly, the CFD (computational fluid dynamics) simulation analysis method was used to analyze the generation mechanism of pressure-flow pulsation at the upper/lower dead points of the axial piston pump; secondly, the influence of damping groove structural parameters on the output pressure-flow pulsation of axial piston pump was analyzed, and a multi-objective optimization model of damping groove structure was constructed; finally, the structure of the low pulsation damping groove was solved. The optimized structural parameters were as follows: the damping groove radius was 2.21 mm, the damping groove length was 10.32 mm, and the damping groove deflection angle was 16.54°. After optimization, the pressure pulsation rate was 0.59%, which was reduced by 0.16% compared to the pre-optimization value of 0.75%, and the pulsation amplitude was 0.25 MPa. The flow pulsation rate was 12.02%, which was reduced by 43.59% compared to the pre-optimization rate of 55.61%. The research results provide effective theoretical support and practical guidance for the optimal design of low pulsation structure of axial piston pump.

Fused deposition modeling (FDM) 3D printing requires the print nozzle to be heated the desired temperature of the material before printing begins. Due to the low printing efficiency of the single nozzle FDM 3D printer, and the large lag and poor stability of its heating system, the whole forming process is time-consuming and wasteful of resources, and the quality of the formed parts is not high. In order to solve the above problems, a temperature control method based on genetic algorithm-fuzzy PID (proportional-integral-derivative) was proposed to control the heating method of dual-nozzle FDM 3D printer, which combined the differences in physical and chemical properties of printing materials. The MATLAB/Simulink simulation model of the temperature control system was established to verify the reliability of the proposed control method. The simulation and experimental results showed that compared with the traditional PID control and fuzzy PID control, the response time of the genetic algorithm-fuzzy PID control was shortened by 36.03% and 32.45%, and the adjustment time was shortened by 28.06% and 20.99%, which had the advantages of fast response, short adjustment time, small overshoot and stable control effect. The research results can provide reference for dual-nozzle FDM 3D printing of composite materials.

Aiming at the problems of poor forming quality and long printing time of concrete 3D printing components, a path planning algorithm based on continuous vertex partitioning was proposed. Firstly, the continuous vertex partitioning method based on Hamiltonian circuit was used to divide the print area into several continuous regions to ensure that the print nozzle would not pass the same vertex many times during the printing process, thus avoiding the problem of repeated printing and poor forming quality. Then, the genetic algorithm was used to search each region, and the shortest printing path was determined through iteration and optimization. The experimental results showed that compared with other path planning algorithms, the proposed algorithm could significantly reduce the empty travel and start-stop times of the print nozzle, and shorten the printing time by more than 10%, which effectively improved the forming quality and printing efficiency for concrete components. The concrete 3D printing path planning algorithm based on continuous vertex partitioning solves the problems of poor forming quality and long printing time of concrete components by effectively dividing the print area, intelligentiy searching the shortest path and combining the optimal path, which can provide strong technical support for the development and application of concrete 3D printing technology.

Aiming at the problems of large changes in longitudinal pitch amplitude and longitudinal pitch angle caused by the flow field when the remotely operated vehicle (ROV) sails at high speed, a method for achieving high speed underdriven ROV operated with zero longitudinal pitch or slight longitudinal pitch by matching and selecting structural parameters of the extended chassis and the tailplane was proposed. Based on the lattice Boltzmann method (LBM), a six-degree-of-freedom simulation experiment was carried out by using the wall adaptive refinement algorithm combined with the structural parameters of the ROV to simulate the ROV sailing motion. The numerical analysis for the ROV with different height of extended chassis and tailplane was analyzed numerically to obtain the relationship between the structural parameters of extended chassis and tailplane and the longitudinal pitch amplitude and longitudinal pitch angle. By comparing the rotating torque of ROVs with similar sailing performance, the relationship between ROV stability and tailplane height was determined under the same extended chassis conditions. The orthogonal experiments for structure optimization of the extended chassis and tailplane were conducted, and the longitudinal pitch data of the ROV under different experimental schemes were fitted by using genetic algorithm. Combined with the actual requirements, the height of the extended chassis and tailplane was determined, and the correctness of the ROV longitudinal pitch design scheme was verified by the actual test. The results showed that the reasonable match of the extended chassis and tailplane structure could effectively reduced the longitudinal pitch amplitude of underdriven ROVs, so as to achieve slight longitudinal pitch sailing motion of ROVs without amplitude compensation. The research results can provide reference for improving the longitudinal pitch motion of relevant underwater devices.

The construction of supporting wheel-track-ground multi-body coupling system is an important link for the travelling smoothness study of tracked robots. Taking the unilateral five- supporting wheel tracked robot as the study object, the theoretical estimation model for tracked robot vibration was established on the basis of the nonlinear suspension system model and the road excitation considering the track filtering effect. Then, based on the simulation experiments of the tracked robot under working conditons with different road levels and different driving speeds, the effects of driving speed and road excitation on the vertical vibration of the robot body and its supporting wheels were quantitatively analyzed by the root mean square value. Finally, the theoretical vibration estimation model and dynamics simulation model of the seven-degree-of-freedom semi-vehicle were verified by external road experiments. The results showed that the vertical vibration of the tracked robot body increased linearly with the increase of driving speed and road roughness, and the track had a filtering effect on the road excitation. The vertical vibration acceleration of the tracked robot body centroid and the supporting wheel 1, 3, 5 were most sensitive to the road excitation, and the power spectral density amplitude of the vertical vibration acceleration of the robot body was the largest when the frequency was about 19 Hz. As the supporting wheel 5 was close to the driving section, the polygonal effect generated by the track engaging with the active wheel made its vertical vibration larger than that of the other supporting wheels. The combination of theoretical modeling, simulation analysis and external experiment provides a new idea for the study of vibration response characteristics of tracked robots under different road conditions.

In order to realize the multi-modal motion of continuum robotic arm and solve the problem that the existing robotic arm can only achieve single bending or rotation, a single-module multi-degree-of-freedom flexible continuum robotic arm based on rolling contact is designed. The rolling contact module was used as the skeleton structure of the bending module in the continuum robotic arm, and the rotating module was installed in the bending module to form a multi-degree-of-freedom robotic arm with independent bending and rotation motions. The kinematics model of the continuum robotic arm was established by the segmented constant curvature method, and its stiffness, bending and rotation properties were analyzed. A continuum robotic arm prototype was prepared, and experiments were carried out on the robotic arm to unscrew bottle caps, switch on the fan, and grasp the object by avoiding obstacles in the three-dimensional space. The experimental results showed that different tasks in complex spatial environment could be accomplished by the combined motion of bending and rotation, which reflected the advantages of the composite motion mode. The designed continuum robotic arm has multi-modal motion, which provides a new idea for the design of multi-degree-of-freedom continuum robotic arms and expands the application scenarios of continuum robotic arms.

Conventional robotic hands usually require drives to continuously provide torque or force to maintain grasping state. If the drive fails, the robotic hand cannot grasp the object stably. To solve these problems, a passively-triggered bistable robotic hand based on origami is proposed. The robotic hand was composed of a grasping mechanism with single-degree-of-freedom and a driving mechanism with bistable characteristics. Based on the kinematics models of the grasping mechanism and driving mechanism, the structure of robotic hand was designed according to the requirements of grasping state, and the energy barrier could be adjusted flexibly by setting the stiffness parameters of the torsion spring. Finally, the drop capture experiments were carried out to verify the grasping performance of the designed robotic hand. The results showed that when the drop height was 400 mm, the grasping motion of the robotic hand was not triggered; When the drop height was 440 mm, the robotic hand successfully grasped the object; When the drop height was 480 mm, the robotic hand failed to grasp the object although the grasping motion was triggered. The experimental results not only validate the ability of robotic hand to grasp objects of a certain size stably without driving, but also reveal the existence of energy barriers under different external shocks. The passive-triggered bistable robotic hand based on origami has potential applications in passive and adaptive robots.

At present, the 3D printed concrete field is still hampered by numerous issues that impede its large-scale industrial production and application. Among these, pores stand out as the most prevalent defect. Consequently, there is an urgent imperative to develop pertinent detection technologies for enhancing the printing quality of concrete. Aiming at the existing 3D printed concrete interface pore detection methods that mainly rely on subjective experience of individuals, and have disadvantages such as long-time consumption, high cost and large computational resource consumption, a lightweight intelligent pore detection method is proposed by introducing a deep learning-based object detection algorithm. Firstly, the traditional image processing algorithms were employed to preprocess the 3D printed concrete interface pore images, and the pore image dataset suitable for the target detection algorithm was constructed. At the same time, based on the characteristics of the constructed dataset, the anchor-box calculation method was optimized to acquire anchor boxes that were better suited to the interface pore targets, so as to improve the detection accuracy. Then, within the backbone network of the detection method, the ShuffleNetv2 network was utilized for multi-scale feature extraction, and part of the network was removed to reduce the network depth and the number of calculation parameters, thereby enhancing the pore detection efficiency. Finally, in order to improve detection precision, the polarized self-attention mechanism module was incorporated into the feature extraction network to enhance the attention to the pore target while maintaining high resolution. Experimental results demonstrated that the proposed method could effectively complete the intelligent detection of 3D printed concrete interface pores. Through comparing with various detection algorithms, it was found that multiple performance indicators of the method were improved, and the detection efficiency was significantly boosted. The research results can provide some data support for the subsequent quality control and performance evaluation of concrete.

The water-cooled wall of boiler in thermal power plants needs to be inspected and cleaned regularly. Using water-cooled wall robot can improve the efficiency of inspection and cleaning. In view of the complex working environment of water-cooled wall, a new type of water-cooled wall robot was developed. The robot structure and working principle were introduced. In order to ensure the robot to move flexibly and have reliable suction on the water wall, an electric permanent magnet wheel was designed. Through Maxwell simulation and experiment, the current excitation required to magnetize/demagnetize the electric permanent magnet wheel and wheel suction were obtained, and the electric permanent magnet wheel magnetize/demagnetization circuit was designed. The control system of the robot body was introduced, and the experimental platform for lateral walking of the robot was built to verify the cooperation and stability of the robot motion. The experimental results showed that the inner and outer legs of the robot could adsorb and move forward alternately, and realize the gait of lifting, stepping and dropping legs, and the movement was stable. The magnetic force was produced when the robot droped its legs and disappeared when it lifted its legs. The robot had both adsorption stability and movement flexibility. The electric permanent magnet wheel had simple structure, small size, small mass, less power consumption, and could provide about 150 N suction. The research results provide a reference for the application of wall-climbing robot in the cleaning and detection of water-cooled walls.

As a vital component of the magnetic adsorption wall-climbing robot, the structure of the magnetic adsorption module usually affects the overall mass and adsorption stability of the robot. Aiming at the problems of complex magnetic circuit coupling relationship and complicated optimization design of magnetic adsorption modules, a magnetic adsorption module structure optimization method is proposed by combining virtual simulation technology, surrogate model and dung beetle optimization algorithm to improve the efficiency of magnetic force calculation and optimization design process. Firstly, the structure design scheme for the wall-climbing robot was introduced, and through the simulation analysis of the existing Halbach array magnetic circuit modes, it was determined that the three-magnetic circuit mode had relatively high adsorption efficiency. At the same time, the magnetic force simulation model of the magnetic adsorption module was experimentally verified based on the initial parameters, which laid the foundation for establishing subsequent surrogate models. Then, an optimization model with the robot's adsorption stability and structural parameters as constraints and the lightweight of the magnetic adsorption module as objective was established. A fourth-order response surface model between the magnetic force and the structural parameters of the magnetic adsorption module was established by the optimal Latin hypercube design, ANSYS parametric modeling and surrogate model technology, and its credibility was verified. The structural parameter optimization model of the magnetic adsorption module was solved by using the dung beetle optimization algorithm. The results showed that the prediction error of the established surrogate model was tiny, and the relationship between the magnetic force and the structural parameters of the magnetic adsorption module could be well expressed. After optimization, the mass of the magnetic adsorption module was reduced by 12.7%. Finally, the correctness of the optimization process was verified through robot load experiments. The research results can provide reference for the magnetic force analysis and structure optimization of other magnetic adsorption robots.

Soft pipeline robots with anchoring-telescoping motion mechanism are typically constructed from flexible materials such as silicone and hydrogel, which can realize anchoring and telescoping in the pipeline through the deformation of flexible materials and have good flexibility. However, due to the viscoelasticity and hysteresis of flexible materials, the soft pipeline robot usually exhibits small force and slow response speed, which is difficult to store and release a large amount of mechanical energy quickly, and the crawling speed is slow. In order to solve this problem, a soft pipeline robot that can realize fast crawling is designed. This robot was composed of an anchoring module and a telescoping module. The anchoring module employed flexural deformation of the flexible belt to achieve the anchoring in the pipeline, while the telescoping module utilized the soft continuum structure with tower springs as main part to facilitate extension and contraction. According to the experimental measurement, the maximum crawling speed of the robot in the pipeline was 102 mm/s and the maximum anchoring force was 76.4 N. The robot was capable of stable crawling in pipelines with inner diameter of 90-120 mm, and had good adaptability to different shapes of unstructured pipeline environment. The results demonstrate that the designed robot can not only achieve bidirectional crawling in horizontal and vertical pipelines, but also quickly pass through S-shaped pipelines, which can provide new ideas for the design and research of soft robots in unstructured pipelines.

Considering the influence of the uncertain parameters of central pattern generator (CPG) model on the kinematic stability of hexapod robot, an optimization design method for the kinematic stability of hexapod robot based on a probability-interval hybrid model was proposed. Firstly, the numerical model of the hexapod robot was established, and the CPG model of the hexapod robot was established based on the Matsuoka and Kimura models. Secondly, the uncertainty variables of the CPG model were described by the probability-interval hybrid model, and the kinematic stability optimization mathematical model of the hexapod robot was also constructed. Then, Karush-Kuhn-Tucker (KKT) optimization condition and the second order fourth moment method based on the maximum entropy principle were used to decouple the kinematic stability optimization design problem of the hexapod robot, and the three-level nested optimization design problem was transformed into a single-level optimization design problem, which realized the efficient solution of the optimization problem. Finally, the kinematic stability approximate model of hexapod robot was established based on radial basis function, and the optimal design solution was obtained by genetic algorithm. The results showed that the proposed method could effectively solve the optimal parameters of the CPG model and improve the kinematic stability of the hexapod robot. Therefore, this method has high application value in the field of robot motion control.

The repair of large-skin wounds has been a difficult problem to be solved urgently. At present, the commonly used repair methods are mainly autologous skin transplantation and wound dressing treatment, but these methods cannot simultaneously meet the needs of large-skin repair and customized repairment. The in situ skin printing technology provides a new idea for the repair of large-skin wounds. However, the existing bioprinting equipment has small printing range and low printing precision, which cannot realize the shape printing of large area of skin tissue. In order to solve the above problems, an in situ skin printing system composed of Stewart parallel robot, linear module mechanism, print head and 3D scanner was proposed. The Stewart parallel robot could achieve high-precision skin in situ printing as printing driving device due to high repeated positioning precision and low cumulative error. The Stewart parallel robot had six degrees of freedom and could adjust the printing angle in 3D space, allowing bioink to fully cover the skin wounds along the skin surface, which was beneficial for wound repair. In order to analyze the feasibility of the designed in situ skin printing system, the workspace of the parallel robot was calculated by numerical method, and the working range of the in situ skin printing system was obtained and verified through printing experiments. The experimental results showed that the parallel robot operated according to the specified path, and the print head could stably inject bioink during the printing process. The working range of the in situ skin printing system was basically consistent with the workspace of the parallel robot, which met the needs of repairing large-skin wounds. The research results lay a theoretical foundation for the subsequent animal experiments on large-skin repair.

Aiming at the shortcomings of traditional sliding mode control in trajectory tracking of boom-type roadheader, such as slow global convergence and significant chattering, an improved sliding mode control method based on novel reaching law is proposed. By introducing lateral deviation and heading angle deviation of the roadheader body and adding power reaching term to the traditional exponential reaching law, the rapid convergence of trajectory deviation and chattering reduction for the roadheader were achieved. At the same time, the boundary layer method was employed to further suppress chattering, which addressed the problem of chattering easily caused by the product term of sign functions in the reaching law. The existence, reachability and stability of the novel reaching law were analyzed, and the interval of disturbance steady-state error was derived. Considering the uncertain disturbance of the roadheader, the simulation comparison was conducted between traditional sliding mode control method and improved sliding mode control method. The results indicated that the control accuracy, convergence speed and anti-interference ability of the improved sliding mode control were superior to the traditional sliding mode control. Finally, an experimental platform was set up to test the performance of the roadheader trajectory tracking control system, which verified the feasibility and effectiveness of the improved sliding mode control method. The research results can provide important reference for the intelligent control of mining equipment in the harsh environment of underground coal mine.

In order to better assist the rehabilitation training for hemiplegic patients, a lower limb rehabilitation exoskeleton robot driven by disk motor is designed, and the effectiveness of its different rehabilitation training modes is verified through visualization research of power-assisted effect and performance analysis. Firstly, the detailed structural design for the lower limb rehabilitation exoskeleton robot was performed, and the biomechanical analysis of human-machine coupling was carried out by using OpenSim software. Then, the passive rehabilitation training experiment based on position tracking control and resistance rehabilitation training experiment were carried out, and the surface electromyographic signals were collected to verify the effectiveness of the designed lower limb rehabilitation exoskeleton robot to assist patients in rehabilitation training under different modes. The results showed that wearing lower limb rehabilitation exoskeleton robot could reduce the human knee joint torque by about 50%. In the passive rehabilitation training experiment, the following error was within -4°-8°, and the muscle activation of the target muscle group of human lower limbs showed an obvious periodic change. In the resistance rehabilitation training experiment, the muscle activation of the target muscle group of human lower limbs increased with the increase of weight. The designed lower limb rehabilitation exoskeleton robot has good sensitivity and followability, and its passive and resistance rehabilitation training modes are conducive to lower limb rehabilitation of hemiplegia patients, which has broad application prospect.

In order to solve the problem of difficulty in testing and performance verification of compliant mechanisms in the process of optimal design, the mechanical properties of thermoplastic polyurethane (TPU) material were studied by 3D printing technology. The effects of material hardness and print fill rate on the mechanical properties of TPU material were analyzed, and the better 3D printing parameters of TPU material were obtained. Using single factor and two factor tset combined with variance analysis, the primary and secondary factors that significantly affect the flexibility of TPU specimens were identified as TPU material hardness and print fill rate. Combined with the mechanical property test data of TPU material, the mapping relationships between the material parameters and material hardness, print fill rate of four commonly used hyperelastic material constitutive models, i.e. Mooney-Rivlin, Yeoh, Ogden and Valanis-Landel, were obtained. The results showed that with the increase of TPU material hardness and print fill rate, the flexibility of the specimens decreased; among the four hyperelastic models, Ogden model has a good prediction effect on the mechanical properties of TPU specimens under different printing parameters; there was no significant difference in the predictive effect of the four models under the same TPU material hardness and different print fill rates. The research results can provide reference for 3D printing and finite element simulation analysis of TPU material, and provide reliable technical support for the test, performance verification and sample production of compliant mechanisms in the design process.

Multi-functional small robots have broad application prospects. To meet different operational requirements, a small land-air deformable amphibious robot was designed, which could achieve efficient ground movement and avoid obstacles through takeoff flight. The robot adopted a dual-mode design, in which the ground mode adopted a two-wheel drive motion design, and the airplane mode adopted a four-rotor flight design. The switching between the two modes was realized through the support and extension of the robot tilting mechanism. In order to verify the motion performance of the robot, the whole robot model was established by SolidWorks software, the kinematics modeling for the robot was carried out, and the kinematics equation of the robot mode switching process was derived. Then, the robot servo output was simulated by MATLAB and the robot prototype mode switching experiment was carried out. The simulation results of the output torque were basically consistent with the measured results, with a range of 0-250 N·cm. Finally, the robot prototype was used to conduct ground movement and air flight tests, and its motion process was analyzed to verify the stability of the land-air motion and mode switching of the robot. The research results verify the effectiveness of the designed robot, and it has a long endurance, which can provide a reference for the design of land-air amphibious robots.

Aiming at the key issues in the overall development of snakelike robots, including material selection, structure design and motion realization, a new multi-joint snakelike robot was developed. This snakelike robot was composed of 11 two-degree-of-freedom orthogonal joints, which could achieve three-dimensional high biomimetic motion while ensuring flexibility. The basic gaits of snakelike robot such as meandering, wriggling and tumbling were designed by using the serpentine curve, and an improved obstacle surmounting gait was further proposed. At the same time, the gaits of the snakelike robot were simulated in the V-REP software, and the motion trajectories and efficiency of different gaits were compared. Finally, through the gait experiment of the snakelike robot prototype, the influence of each control parameter in the gait model on the motion waveform and speed of the snakelike robot was analyzed, and the reliability of the body structure and control system of the snakelike robot was verified. The research results have important theoretical significance and practical guiding value for realizing the gait planning and motion control of snakelike robots.

In order to solve the problems of poor wall adaptability and low movement flexibility of wall-climbing robots, the shortcomings of the existing magnetic adsorption mechanism of wall-climbing robots were analyzed. Taking a wheeled wall-climbing robot as research object, a wall adaptive pendulous magnetic adsorption mechanism was designed based on the functional requirements of wall-climbing robots. A comparative analysis was conducted between the pendulous magnetic adsorption mechanism and the traditional magnetic adsorption wheel by Ansoft software. In order to further reduce the mass of the magnetic adsorption mechanism and improve its adsorption reliability, based on the goal of high magnetic energy utilization, SNLP (sequential non-linear programming) algorithm was used to optimize the structural parameters of the pendulous magnetic adsorption mechanism. After optimization, the adsorption force of the magnetic adsorption mechanism was increased by 25.52%. A prototype of pendulous magnetic adsorption wheel was developed and the adsorption force testing experiment and demagnetization experiment were conducted. Motion performance testing experiment was carried after installing the pendulous magnetic adsorption wheel on a wall-climbing robot to verify the rationality of the optimization results of the structural parameters of the magnetic adsorption mechanism and the practicality of the structural design. The research results provide a reference for improving the working performance of wall-climbing robots.

As an important spatial force sensing element, the six-dimensional force sensor is widely used in robot force/position control, grasping assembly, contour detection, autonomous obstacle avoidance and human-computer interaction. At present, improving the accuracy is one of the main research directions of six-dimensional force sensors. However, due to the influence of own structure and processing error and other factors, the six-dimensional force sensor will produce the interdimensional coupling phenomenon, and the interdimensional coupling is an important factor affecting its accuracy. In order to reduce the influence of coupling error, the decoupling algorithm of six-dimensional force sensor is studied by combining error analysis, theoretical derivation and experimental verification. Firstly, the coupling analysis of the six-dimensional force sensor was carried out, and its coupling model was obtained. Then, the linear decoupling algorithm of the six-dimensional force sensor was studied, and on this basis, the decoupling algorithm based on polynomial fitting was proposed to reduce the coupling error without changing the structure of the six-dimensional force sensor, so as to improve its accuracy. Finally, the orthogonal parallel six-dimensional force sensor was selected to carry out calibration experiments, and two algorithms were used for decoupling solution. The results showed that the decoupling algorithm based on polynomial fitting could reduce the influence of interdimensional coupling on the accuracy of six-dimensional force sensors. The proposed decoupling algorithm effectively improved the accuracy of the six-dimensional force sensor. Compared with the linear decoupling algorithm, the maximum coupling error was reduced by 8.914 percentage points and the linearity error was reduced by 0.111 percentage points. The research results can provide reference for reducing the coupling error and improving the accuracy of six-dimensional force sensors.

In response to the issue of target disappearing when a mobile robot is following a target, a robot target following system based on visual tracking and autonomous navigation is proposed. The robot following problem was divided into two cases: regular following when the target was within the robot's field of view, and autonomous navigation after the target disappeared. For the former case, the target's motion state was predicted using a Kalman filter, appearance features were extracted using a pedestrian re-identification network, and target tracking was performed by fusing motion information and appearance features using data association fusion. Servo control was then applied for following the target. For the latter case, an autonomous navigation algorithm was adopted based on the relative position between the historical target and the robot. The robot moved to the history position of the target and searched the target, aiming to increase the success rate of the target following. Evaluations were conducted on the OTB100 benchmark dataset and a target following test dataset which was in robot application scenarios. Experiments were performed on a mobile robot platform. The results showed that the robot could follow the target in the environment with different lighting conditions and more background pedestrians, which verified the robustness and effectiveness of the proposed algorithm, and it could meet the real-time requirement. The research results can provide reference for research on the problem of robot refollowing after the target disappears.

In order to further improve the response speed and lightweight extent of piezoelectric wavefront correctors, the dynamics simulation and optimization method of unimorph piezoelectric deformable mirrors for high dynamic shape control was studied. Firstly, based on the parameterized finite element model of unimorph piezoelectric deformable mirror, an efficient and high-precision electromechanical coupling dynamics simulation analysis method was proposed. Then, through orthogonal traversal of multi-dimensional design parameters (material type, geometric dimension and fixation mode of the optical reflector and piezoelectric ceramic), the influence of different parameters on the dynamic shape control performance of unimorph piezoelectric deformable mirrors was explored. Finally, the optimized design scheme of the unimorph piezoelectric deformable mirror with expected response bandwidth and actuation displacement performance was obtained. The experimental results show that the natural frequency and mirror actuation displacement amplitude of the developed unimorph piezoelectric deformable mirror are in line with expectations, which verifies the effectiveness of the proposed dynamics simulation and optimization method and provides a scientific theoretical method for the efficient development of high-bandwidth unimorph piezoelectric deformable mirrors.

Aiming at the shortcomings of rigid robot arm in fruit and vegetable picking, a flexible picking robot arm with simple structure and flexible movement was designed. Firstly, the kinematics model of the flexible picking robot arm was established based on the equal arc hypothesis, and the forward and inverse kinematics analysis from joint space to operation space and from driving space to joint space was carried out, as well as the decoupling analysis between the flexible joints. Then, the kinematics models of the flexible picking robot arm were numerically calculated by using MATLAB software, and the virtual prototype model of the flexible picking robot arm was established by ADAMS software. The kinematics simulation was carried out under the same working conditions as the theoretical analysis, and the accuracy of the theoretical analysis results was verified. The simulation results showed that the flexible picking robot arm could move flexibly and coordinatedly. The research results can provide a basis for the subsequent motion control of flexible robot arms.

In order to solve the problems of low automation level, low work efficiency and high risk when launching and recovering autonomous underwater vehicle (AUV), an launch and recovery system (LARS) for AUV based on unmanned surface vessel (USV) was developed. Firstly, by analyzing commonly used AUV launch and recovery modes and LARS at home and aboard, an AUV LARS was designed, and its working principle was analyzed; secondly, the dynamics, statics and contact collision issues of the LARS were studied from the perspectives of mechanical analysis and numerical simulation; finally, a principle prototype of the system was built and land and lake experiments were conducted. The experiments verified that the designed LARS was stable, reliable, easy to operate, and had good universality, which could effectively improve the efficiency of autonomous launch and recovery of AUV. The designed AUV LARS based on USV has good application prospects.