Intelligent, handy berry harvester with AI ripeness detection and soft gripper

Abstract

A hand-held berry harvesting device consisting of a ripeness detection module with sensor arrays for assessing color and firmness, a microcontroller for real-time analysis of sensor data using an artificial intelligence model trained on ripeness thresholds to identify ripe berries, a soft, adaptive gripper for enclosing the identified berries with adjustable pressure, a low-frequency vibration module to assist detachment from the plant, and a collection path for receiving the harvested berry.

Description

AREA OF INVENTION

[0001] The invention relates to agricultural harvesting devices with integrated sensor arrays for determining ripeness, integrated AI analysis, soft robot grip and controlled low-frequency shedding for delicate berries under field conditions. BACKGROUND OF THE INVENTION

[0002] The selective manual harvesting of delicate berries remains labor-intensive and prone to errors. Visual ripeness indicators and variable stem holding forces lead to inconsistent quality and damage. Automation approaches using computer vision, tactile / rigid sensors, and soft grippers to reduce crushing and tearing are promising. Soft robot end effectors with compliant fingers, cable drives, and integrated sensors have already been used for blackberries and similar fruits. They combine spectral or visual ripeness detection with a gentle gripping mechanism to improve detachment without crushing. Non-destructive firmness and tactile sensing methods, including compliant grippers with integrated sensors, enable the classification of fruit firmness and ripeness, complementing color / reflectivity information in dense canopies with varying light conditions.Detachment by low-frequency vibration or shaking is a known technique for increasing the harvest of ripe fruit while minimizing the waste of unripe fruit, provided the frequency and stroke are adjusted. This suggests that handheld devices can utilize controlled vibrations to assist separation at the stem with minimal damage. Under German utility model patents, product-oriented claims to devices and systems are permitted, while process claims are generally excluded. This favors a device formulation that incorporates the described work steps as functional modules. SUMMARY OF THE INVENTION

[0003] The invention relates to an intelligent, handheld berry harvesting device comprising the following components: a ripeness detection module with sensor arrays for determining color and firmness; an integrated microcontroller executing an AI model trained on ripeness thresholds; a soft, adaptive gripping module with adjustable pressure; and a low-frequency vibration module to assist detachment. All components are integrated into an ergonomic housing with a collection trough and optional data acquisition. The device scans target berries, analyzes sensor data in real time to identify ripe berries, grips a selected berry with compliant grippers under controlled force, applies calibrated vibrations to detach the fruit from the plant, and releases it into a collection area while simultaneously recording harvesting data.Preferred embodiments utilize a cable-operated, soft gripper with compliant silicone fingers and integrated force / pressure sensors, multispectral or RGB-NIR optics for determining ripeness color indices, and a small inertial actuator for low-frequency vibrations with amplitudes and frequencies optimized for harvesting a high yield of ripe fruit with minimal loss of unripe fruit. DETAILED DESCRIPTION

[0004] The compact housing features a forward-facing sensor head that combines an imaging module (visual or multispectral) and, optionally, tactile / firmness measurement via a flexible probe or sensors integrated into the gripper. This allows for the detection of ripeness characteristics, insensitive to lighting conditions and occlusion. Reflectance indices and texture are calculated internally to determine the degree of ripeness. An integrated microcontroller runs an AI model trained on datasets of ripe and unripe berries, utilizing colorimetric data, spectral ratios, and tactile / firmness characteristics. The model outputs include berry classification and safety, and control subsequent gripping and release parameters. The soft gripping module consists of multiple flexible fingers made of elastomeric materials, actuated by either a cable or pneumatic mechanism.Finger geometry and stiffness are designed to adapt to the berries while distributing contact pressure. This minimizes crushing, as confirmed by studies using soft grippers in fruit harvesting. Force or pressure sensors integrated into the fingertips (e.g., thin force-measuring films) allow for gripping force control and adaptive compliance that adjusts to berry size and firmness. This ensures the gripping force remains below the pressure point threshold relevant for delicate products. A low-frequency vibration module, such as an eccentric or a voice coil actuator, transmits the vibration to the gripper or a nearby structure. The firmware selects frequency and amplitude ranges (approximately 10–14 Hz with a suitable stroke) that promote the release of ripe fruit while minimizing the loss of unripe fruit. It also limits the number of cycles to prevent fatigue.The device features an ergonomic handle with a release mechanism and a small display / LED indicator for ripeness, gripping force, and readiness for release. The collection path positions a chute or container below the gripper for immediate and gentle placement. Power is supplied by a battery designed for single-use field work. Energy management switches off power-intensive modules between harvesting operations. The sealed construction and replaceable fingertips ensure hygiene and durability during outdoor harvesting. The control unit logs harvesting metadata such as timestamps, classification accuracy, release settings, and optional GPS tags for traceability. The data can be exported for yield mapping and quality analysis.Safety features limit the maximum gripping force and the number of release cycles per berry, warn the operator of low-accuracy classifications, and include an emergency stop switch. This compensates for variations in stem holding forces and crown structure. The modular architecture allows for the exchange of sensor heads (e.g., spectral filter sets) and finger geometries for different berry varieties and supports firmware updates to optimize AI thresholds and vibration release profiles based on field data.

Claims

[1] A hand-held berry harvesting device comprising a ripeness detection module with sensor arrays for assessing colour and firmness, a microcontroller for real-time analysis of sensor data using an artificial intelligence model trained on ripeness thresholds to identify ripe berries, a soft adaptive gripper for enclosing the identified berries with adjustable pressure, a low-frequency vibration module to assist detachment from the plant, and a collection pathway for receiving the harvested berry. [2] Device according to the preceding claim, wherein the soft, adaptive gripper consists of elastic fingers with tendon actuation and integrated force sensors for closed pressure control to minimize crushing when gripping and releasing delicate fruits. [3] Device according to the preceding claim, wherein the low-frequency vibration module applies vibrations in a frequency and amplitude range selected to increase the removal of ripe fruit while limiting the detachment of unripe fruit, the number of cycles being limited to reduce fatigue effects. [4] Device according to the preceding claim, wherein the ripeness detection module comprises a multispectral imaging unit and tactile sensors for calculating color indices and strength characteristics, and the control unit logs harvest data including classification certainties and detachment parameters for traceability.