The Problem We're Solving
The global food system is at a critical juncture, strained by urbanization and limited arable land. While indoor and vertical farming present a promising future—offering higher yields and nutrient density with up to 90% less water than traditional methods —they face significant barriers to large-scale adoption. The industry's growth is consistently hindered by two primary factors: prohibitively high upfront facility costs and crippling labor costs for operation and harvesting
​
​
This "scaling problem" was highlighted by the struggles of ventures like AppHarvest, which, despite massive facilities, cited "labor and productivity challenges" as a key reason for its ultimate bankruptcy. The market has ample room for growth, but a technological barrier prevents indoor farming from reaching its full potential. Kurtz Robotics AGROBOT Division was founded to bridge this gap.
Precision Harvesting, Automated.
AGROBOT T.O.M. is our solution to automate the demanding task of harvesting soft produce, beginning with greenhouse tomatoes. This autonomous mobile robot is designed to navigate along crop rows, use advanced sensors to identify perfectly ripe fruit, and gently pick them with a custom soft gripper. By automating this process, we directly address the high labor and facility costs that have challenged the indoor farming industry, offering a modular and efficient path to scalability.
​
Project Goals:
-
Design and build an autonomous mobile robot for greenhouse tomato harvesting.
-
Develop a robust 3D perception system to map plants and locate tomatoes using sensor fusion.
-
Implement computer vision algorithms to accurately assess tomato ripeness and detect damage.
-
Integrate a novel soft robotic gripper for delicate, damage-free picking.
-
Achieve reliable navigation along crop rows using fused odometry data.
A Look Under the Hood
Advanced 3D Perception
Our system uses a combination of LiDAR for navigation and spatial awareness and an RGB-D camera (like the Intel RealSense) as its primary perception sensor. The system fuses data from RGB, LiDAR, and Odometry to build a robust 3D map of the environment and locate tomatoes. A pre-trained AI model processes the camera's color stream to identify ripe tomatoes with 2D bounding boxes. We then project these 2D detections into 3D space by aligning them with corresponding depth data, using a robust median-filtering method within the bounding box to ensure accuracy and mitigate the impact of sensor noise or occlusions.
Intelligent Motion & Manipulation
Once a target tomato is localized in 3D, our motion planning software, built on the ROS 2 framework, takes over. Using numerical Inverse Kinematics (IK) solvers like Vortex solver within the CoppeliaSim Environment, the system calculates a collision-free path for the robotic arm. The system builds a detailed 3D occupancy map of its environment from sensor data to avoid collisions with stems, leaves, or other obstacles during manipulation. For the final, precise approach, we recommend using visual servoing, where the robot uses real-time camera feedback to guide the gripper accurately to the target, compensating for any small errors in initial localization.
The Soft Robotic Gripper
A key innovation is our proprietary soft robotic gripper. Given the delicate nature of produce like tomatoes, a traditional rigid gripper can easily cause bruising and damage. Our soft gripper, an adaptation of Professor Doug Holmes' "Kirigami" Gripper, is designed to handle produce gently, ensuring it can be harvested without compromising its quality, thereby reducing waste and maximizing yield. Using FEA Analysis, our team was able to tune the gripper to exert a constant 20N of force across the entire range of motion of the gripper, allowing for consistent grip profiles among a large range of tomato diameters.
Forged Through Real-World Challenges
SHOULDER V1
AGROBOT - 0001
Motor Controller Bay
AGROBOT - 0001
Back Pitch Brace
AGROBOT - 0001
Shoulder Assembly Test
AGROBOT - 0001
Developing a sophisticated robotic system is a journey of persistence and iteration. Our path was marked by significant technical hurdles that provided invaluable lessons. We overcame challenges including faulty motors that caused delays, difficulty in precisely tuning motors for joint angle control, persistent wiring issues that impacted system reliability, missing or mismatched components during assembly, and complex system-level integration bugs.
​
Key Learnings:
​
-
Test Early and Often: We learned that identifying hardware issues early in the development cycle saves significant time and prevents costly overruns later.
-
Organized Wiring is Essential: Investing in proper cable management, clear labeling, and robust connections pays significant dividends in system stability and ease of troubleshooting.
-
Flexibility in Design: Incorporating flexibility from the outset, such as designing for alternative components, helps mitigate the impact of unexpected parts shortages and supply chain disruptions.
-
Hands-on Iteration Beats Theory: We found that while theoretical planning is essential, hands-on iteration is more effective when dealing with real-world hardware, as some issues only become apparent during implementation and testing.
-
Team Collaboration is Key: Open communication and fostering a collaborative environment, where team members feel comfortable sharing information and seeking help, proved indispensable in overcoming the many blockers we faced.
From Grow Bed to Global Impact: Our Vision for Scalable Farming
Agrobot T.O.M. is the first step. Our long-term vision is to create fully autonomous, modular farming units housed within shipping containers. These units will be deployable anywhere in the world, creating a decentralized and resilient global food system.
Future Unit Features:
-
Fully Autonomous: Each unit will run almost entirely on its own, housing 3-4 full vertical stacks of plants. An integrated robotic gantry will travel the entire unit, monitoring plant health and harvesting multiple crop varieties.
-
Climate Controlled: A full climate conditioning suite will allow the units to operate in any environment, from urban centers to remote regions.
-
Modular and Connected: Units can operate standalone or be chained together [User Query]. A single "base station" will house the central computer and a refrigerated fulfillment dock, allowing for massive scalability with reduced overhead.
-
Data-Driven: The entire system will map the farm in real-time, tracking every single plant to optimize growth, nutrient delivery, and harvest schedules.
Growing a Better Future, Together
Kurtz Robotics is at the forefront of the next agricultural revolution. We are seeking partners and investors who share our passion for creating a sustainable and food-secure future. Join us as we refine our technology, scale our operations, and deploy our vision across the globe.
We are currently seeking:
-
Funding to accelerate R&D, develop our containerized prototypes, and expand our engineering team.
-
Strategic Partnerships with agricultural experts, commercial growers for pilot programs, and manufacturing firms with expertise in robotics and fabrication.
-
Access to Testing Environments to further validate and refine our systems under real-world conditions.
Ready to be a part of the future of food?