| The rapid development and widespread application of plant factories have resulted in a significant reduction in agricultural workforce.Consequently,the adoption of automated picking robots for harvesting is emerging as the dominant trend in modern smart agriculture.In this context,the research and development of cherry tomato harvesting robots for plant factories holds considerable importance in advancing the automation and intelligence of the fruit and vegetable industry.This study focuses on cherry tomato harvesting within a plant factory,specifically concentrating on the functional realization and testing of the hardware system design of the picking robot,the structural design of the picking mechanical gripper,non-destructive fruit picking methods,and the motion planning of fruit picking tasks.The main research and outcomes are as follows:(1)An analysis of the working environment characteristics within plant factories is conducted,considering variables such as the work site,breeding rack and working object.The construction of a Pearson model for "Hongyan H1" cherry tomatoes and a correlation study examining the fruit’s fundamental physical parameters through mathematical analysis.It also examines the compression mechanical properties of various fruit types under different loading directions,explores the impact of different separation positions on fruit detachment from the plant,and compares and analyzes friction mechanics between three different materials and fruits.These results offer crucial theoretical data for the hardware system design of cherry tomato picking robots and non-destructive picking methods.(2)Design of a hardware system for a cherry tomato picking robot tailored to the plant factory environment.This includes the design of a dual-drive chain transmission mechanism for the four-wheel mobile platform,the gear transmission mechanism for the rotating platform with external tooth slewing bearing and spur gear meshing,the ball screw transmission mechanism for the lifting platform with mast structure,and the picking manipulator.The study constructs a model for antioverturning stability,using torque equilibrium methods to study the relationship between the number of masts,mast layout,operational conditions,and the robot’s resistance to overturning,optimizing the design to enhance overturning performance.These findings provide essential hardware support for subsequent research.(3)Based on the structural characteristics,motion characteristics and capture methods of the phalanx chain of vulture claws,two different types of the bio-inspired tarsus end-effector were designed.The RSM-BBD method was used to construct an empirical model for optimizing the structural parameters of the bio-inspired tarsus compliant end-effector,and the optimal structural parameters of the bio-inspired tarsus compliant end-effector were obtained.Subsequently,the key structural parameters of the bio-inspired tarsus rigid single toe were designed by enveloping method.The synchronicity of the bio-inspired tarsus rigid single toe movement with different numbers of toe bones was studied by kinematic simulation methods.At last,the optimal scheme of the bio-inspired tarsus end-effector was determined by fruit picking experiment.The above results provide a hardware foundation for non-destructive picking methods.(4)A non-destructive fruit picking method is proposed to reduce plant damage in response to the problem of mechanical damage during cherry tomato picking.Firstly,the differences between the calibration method of adjusting the resistance and polynomial fitting fusion of the analog-to-digital conversion module and the linear calibration method was analyzed.Then,the relationship between different positions of fruits and the positions,quantities,and loads involved in gripping the phalanges was studied,and the accurate calibration method of fruit surface pressure perception was obtained.Secondly,the evaluation mechanism for non-destructive fruit grasping control was established.Based on this,the effectiveness of non-destructive grasping control method was analyzed by setting safety threshold.Thirdly,the differences in the effects of pulling,bending,and twisting methods on the load of the pedicel and picking time of single fruit were studied.The parameter relationship model between the combination picking method and reducing plant damage was constructed.Finally,the response surface experimental design method was used to study the relationship between the combination picking method and pedicel bearing load and separation force,and the optimal combination picking method was obtained.The above results provide a reference for the motion planning of fruit picking tasks.(5)A motion planning methodology has been introduced,which combines a 43-4-3-4 segmented mixed polynomial spline interpolation with the integration of IDBTRRT and DWLF-PSO.This methodology is specifically designed for the cherry tomato harvesting task within a semi-structured plant factory.Initially,a kinematic model was constructed for the redundant picking manipulator,followed by a comparative analysis aimed at assessing the quality of path planning achieved by the IDBT-RRT algorithm in comparison to similar algorithms across three different environmental complexity scenarios.Furthermore,a study was conducted to examine the impact of four segmented mixed polynomial spline interpolation techniques,including the 3-4-4-3 configuration,on the kinematic parameters of each joint within the picking manipulator.Subsequently,a trajectory planning method utilizing a 4-3-4-3-4 segmented mixed polynomial spline interpolation was proposed.Additionally,taking into consideration critical obstacle avoidance path points and applying a time-optimal optimization criterion,the DWLF-PSO algorithm was employed to refine the trajectory of the picking manipulator.The effectiveness of this algorithm was ultimately validated through a series of simulation experiments.(6)A prototype of a cherry tomato picking robot was developed and validated on a test platform.The results indicate that the maximum static roll angle of the harvesting robot is 26.51°±1.25°.The accuracy of longitudinal deviation,lateral deviation,and angular error were 20.45mm,46.35mm,and 5.11°,respectively.The robot exhibits a maximum movement speed of 1.67m/s,a maximum lifting speed of 0.0897m/s,and a maximum rotation angle of 349.48°.For the picking process,the average time for a single fruit picking operation and the picking action itself was determined to be 14.47s and 6.15s,respectively.The success rate of picking ranged from 85%to 93.94%.In the combination picking method,the picking time for a single fruit was 31.97%longer compared to the pulling picking method.However,this approach resulted in a 1.10%increase in the picking success rate,a significant 72.16%reduction in the maximum separation tension,and a notable decrease in plant and fruit damage rates.Furthermore,it was observed that the IDBT-RRT algorithm can efficiently find feasible paths within specified time in practical scenarios with environmental and task constraints.Leveraging the 4-3-4-3-4 segment mixed polynomial spline interpolation method,the DWLF-PSO algorithm was employed to optimize the kinematic parameters of each joint in the picking manipulator.The outcomes showed that the kinematic parameter curve for each joint exhibited smoothness without abrupt changes.The total average time required for the entire trajectory planning was only 3.4727s,resulting in a 47.06%increase in the working efficiency of the picking manipulator. |
PDF Full Text Request