Design And Experiment Of Intelligent Driving System For Elevated Strawberry Picker

Due to its advantages of good light and ventilation conditions,cleanliness,and labor saving,the planting area of elevated strawberries is constantly expanding.Strawberry picking is the most labor-intensive link in the production of overhead strawberries.The current overhead strawberry picking operation is still dominated by traditional manual picking,which is labor-intensive.The traditional driving and picking method requires people to control the vehicle and pick strawberries with both hands.The manipulation and picking occur at the same time,resulting in low picking efficiency and inconvenient picking and packaging.It is imminent to develop a smart driving system for overhead strawberry picking vehicles.Therefore,this paper aims to develop an intelligent driving system that can be used for both production and leisure agriculture,reducing the fatigue caused by the manipulation of strawberry picking trucks and additional packaging,enhancing the fun of picking and improving operational efficiency.The main research contents of this paper are as follows:(1)The overall scheme of the intelligent driving system of strawberry picking car is proposed.Combining the overhead parameters of the field research with the overhead parameters obtained by the research group in the early stage,observing the operation characteristics of artificially picking overhead strawberries,and putting forward the operation requirements of the strawberry picking truck in the elevated environment.The overall plan includes: the implementation plan of the smart driving system and the interaction plan of the smart driving system.(2)The implementation plan and interaction plan of the smart driving system are designed.In the implementation plan of the intelligent driving system,it mainly includes: the design method of the chassis stable driving module,the foot auxiliary control module and the weighing module.The foot auxiliary control module is used to increase the foot braking function and the foot acceleration function.The stable driving module realizes the stable driving of the chassis,and continuously obtains the quality of the picked strawberries through the weighing module.In the interaction scheme of the intelligent driving system,it mainly includes the design method of picking action recognition module,direction adjustment module,leisure voice interaction module,and voice prompt module.The picking action recognition module recognizes the picking posture of the berry farmers and realizes the behavior and posture control.The motion state of the strawberry picking cart can be adjusted through the direction adjustment module to realize the direction adjustment in the elevated frame.The leisure voice interaction module can realize voice control of the motion state of the strawberry picking cart and the intelligent voice dialogue.The voice prompt module can obtain the strawberry picking quality information sent by the weighing module.Provide voice prompts to reduce labor consumption and strawberry damage.(3)The control system design scheme of the intelligent driving system of the strawberry picking car is proposed.In the hardware system,it mainly includes the key hardware selection of the upper computer system,the lower computer system,the motor drive system,and the visual information acquisition system.At the same time,according to the actual operation requirements,the software of the intelligent driving system of the strawberry picking car is designed,including the design method of the PWM pulse control of the chassis motor,and the communication between the lower computer system and the chassis in the intelligent driving system of the strawberry picking car is proposed.method,the communication method between the upper computer system and the lower computer system,and the DMA data transfer method.(4)Combined with the actual operation situation,a method of picking gesture recognition under the elevated strawberry based on Real Sense D435 is proposed.Through the analysis of picking postures in the actual environment,this paper determines seven picking postures and two picking postures,namely: with picking posture and without picking posture.The collected data is expanded into 1400 sets of pictures,and the data sets of 7 single and double-arm picking postures are classified and marked by Label Img software.At the same time,the field of view of Real Sense D435 is arranged,and a Yolov4-The tiny deep learning model realizes real-time recognition and stop-and-go autonomous operation of picking posture.(5)Five groups of tests were designed in this study,namely: chassis debugging test,weighing and broadcasting test of riding and picking,voice interaction test of riding and picking,attitude control test of indoor riding and picking,and attitude control test of field riding and picking.In the chassis debugging test,when starting work in the soil ground,the soil humidity is less than 60%,there is a small time change,and the chassis performance is stable.In the weighing and broadcasting test of picking by car,the damage rate of picking strawberries by a strawberry picking truck was lower than that of picking by walking.In the voice interaction test of picking and picking,the overall success rate of voice interaction is 80%,which can increase the entertainment in the picking process.In the attitude control test of indoor riding and picking,the identification and control accuracy of the strawberry picking vehicle reached 85%,which can meet the actual picking operation requirements.In the attitude control test of field riding and picking,the parameters were optimized,and when the trigger control confidence was 80%,the overall recognition accuracy of the strawberry picking vehicle reached 93.45%,which met the operation requirements.