Confluence Research and Development, LLC





Let's get to work

Confluence R&D is my design firm. When you work with us on a project, you work directly with me. There are no salespeople, no analysts, no one between you and the person doing the work. Sections of large projects are contracted out, and only to likeminded craftsmen. The final product is unique, customized for you, and built with great care.

The Confluence name comes from experience of successful projects being an engineered mixture of software, hardware, electronics, and science. As an independent contractor, I focus on solutions, for you, undistracted.

As a small firm, the number of projects that can be taken on each year is also small. I only work with people who love using what I create as much as I love creating it. With Midwest values, we only offer what is a good fit, customizing for the perfect fit. If you believe that a small team with pride in their work can deliver something spectacular, let's talk about what you need help with today.

Kent A. Vander Velden, PhD
70/30 Engineer-Scientist
(515) 864-1694
kent@confluencerd.com

Confluence R&D, LLC
6165 NW 86th St
Johnston, IA 50131
USA





Past projects

  • Plant phenotyping instruments (examples below)
  • Deep learning image analysis of plants
  • Telepresence pharmaceutical dispensing
  • Reverse engineering precision agriculture monitors
  • Remote telemetry add-on for precision agriculture monitors
  • Localization platform for autonomous agriculture
  • Handheld spectroscopy of plants and analysis
  • Many analytical software tools
  • Many design proposals




  • Plant phenotyping

    Philosophy

    With over a decade of designing, building, and managing plant phenotyping systems, before offering advice, I first must carefully understand your research goals, and I'll often use that as the foundation of a written proposal. A camera-based plant phenotyping instrument seems to be a box with some lights and a camera, but be careful comparing specifications. To anyone other than a designer, most of the technical specifications will mean little. The differences that matter, alongside cost, is how a choice integrates into your research. I've spent too much time of my career reverse engineering overly expensive instruments to extend their utility or simply make them functional for scientists. An instrument should support the science, not force changes to the science.

    There are options available that provide lots of measurements but supplied as black boxes requiring your experimental protocol to be adjusted.

    We value your time and cost of your experiment. There is nothing worse than as a data scientist discovering that an experiment is unusable because of a data collection mistake. We provide tools to evaluate at the instrument and at the desk, the real-time performance of an experiment being imaged. We want no experimental data discarded.

    We recognize the cost of space. A compromise must be struck between experiment size (more plants are better) and experiment costs (each plant is expensive.) We help by providing analysis methods that are more robust to small experiment sizes.

    We recognize not everyone is a data scientist. Data analysis is full of landmines and data presentation is made difficult because of the breadth of background present in a group meeting. To help, we produce reports that provide alternative explanations using intuitive methods.

    We believe that data quality begins with the operator. The instrument operator has a critical role is data quality, and our software provides immediate feedback of image quality and experiment performance. With other software, the operator is often not able to see the current image without purchasing or running additional software, slowing down collection.

    We do not charge extra for software. We provide all the software that is needed to use the instruments to their potential and are glad to make improvements for you. We don't charge for extra copies, or installations, per image, or any of the other popular customer entrapment methods.

    We despise black boxes. All our analysis software is written in Python. Your data scientists can review and modify the code. Or, we'll update the software for your needs and to integrate into your science.

    We don't up-sell. If you don't require a particular piece of hardware, it'll not be pushed into the instrument customized for you.

    We don't feature-pack. When an instrument produces a lot of metrics, many times the metrics are derived and not fundamental measurements. The uncertainty and vagueness increases.

    We are a family business. No investors. No banks. No pressure to meet quarterly numbers instead of doing what is right. We are family-funded and successful projects promote growth.

    We are a Midwest USA company, proud to serve our domestic customers. We believe this improves communication, focus, and response time.

    Portfolio

    I started designing plant phenotyping hardware after having spent too much time adapting other systems. Instead of fixing an expensive system, with only a fraction of its measurements usable, why not start from scratch and develop systems that do the job that's requested well. Here are a couple of machines from our portfolio, one for the lab and another for the greenhouse, to demonstrate what an individual or a small team can do. After years of service, they are used daily. I'm proud of what has been achieved and how the work is used by plant scientists developing breakthrough traits and treatments. What traits do you want to measure and how would a custom system accelerate your discoveries?

    PlantGazer

    PlantGazer combines three cameras and a rotary platform to photograph plants from several orientations. The cameras move independelty to maximize the detail in each photograph. All images and results are immediately visible to the operator so they may review and make any alterations while still possible. Potted plants up to 18" diameter and 30" high are ideal.


    TrayGazer

    TrayGazer accepts standard 1020 nursery trays with backlighting of Petri dishes as an option. The camera module on top is removable and has taken various forms including: a single stationary camera, single moving camera, or eight-cameras. These options explored the balance between cost and performance. The single camera system can be the least expensive and the eight camera system can be the highest performance, but the moving camera system is ideal for most. The camera moves and takes a photograph from above the center of each tray insert.


    Uniform interface and analytics

    Each instrument has similar appearing and operating software, so it's easy for an operator to switch between machines. Images from any number of our systems are stored to and processed by a workstation that we supply with the first instrument. The images are organized for easy access and processed using machine learning models. For consistancy, the preview results shown at the instrument are processed by the same models. The operator can flag individual plants as appearing particuarlly good or poor for refinement of the model. Model refinements can be automatic or manual. With a little customization, the system can injest images from other sources. Let us know your needs and wishes.




    Testimonials

    It is with great pleasure that I write this recommendation. I have been fortunate to work very closely with Kent, even if for a brief period of time, on a small cross-functional technical team consisting of key statistical geneticists and scientific programmers. The technical task given to the team in question involved translation into scientific software of complex and computationally intensive statistical genetics methodology. Kent's technical excellence, clarity in execution and teamwork have been fundamental in the successful completion of a challenging technical project executed under tight timelines. It is especially his contribution with respect to scientific software code optimization that has resulted in outstanding improvements in terms of software run time execution, with the practical outcome of implicitly enabling wide and fast adoption of key technologies.

    Liviu Radu Totir, PhD
    Biostatistics Lead in Systems and Innovations for Breeding and Seed Products
    Corteva Agriscience



    I have known Kent for a number of years; I was one of his committee members for his Ph.D. He is one of the brightest minds that I have encountered. The powerful aspect to his personality is that he focuses his intellect on solving real world problems that enable people and initiatives to move forward. He combines broad, deep thinking with engineering hands to get work done. I have had more than one experience of asking him to tackle a difficult issue that others could not address, and he solved the problem. His energetic curiosity and ability to learn are engaging for those who want to work on difficult problems in the world today.

    Lane Arthur, PhD
    Director
    Digital Solutions, John Deere Precision Ag



    Kent has been a technology and thought leader in several areas of informatics for many years. His expertise spans from hardware for data acquisition through data processing and analytics to data management and decision support. Scientific programming and optimization as well as data visualization are among Kent's greatest strengths. Kent is creative, in tune with newest technology development, and always focused on pragmatic and robust solutions for business needs.

    Jochen Scheel, PhD
    Director, Data Science and Informatics
    DuPont Pioneer



    I've had the pleasure of working with Kent in a variety of capacities and on a variety of projects during my ten years at Pioneer. Kent is a gifted scientist and engineer, and in my experience is one of the few people who credibly be called by both those names. I've seen him add transformational value to multiple projects at the interface of software/ hardware and biology. His work ethic and passion to deliver is phenomenal and he works across disciplines and with diverse people and problems with ease.

    Jonathan Lightner, PhD
    Vice President, Agricultural Biotechnology
    DuPont Pioneer



    Kent is excellent in analyzing and solving problems. He has tremendous knowledge and skills in a variety of areas. I have worked with Kent on various high performance computing projects over the years and he has been instrumental in their success. He is superior in his ability to take a solution and optimize it for the best performance. I was always excited to work with Kent on a project because of his insight and abilities.

    Kraig Snodgrass
    Senior IM Manager, UNIX/SAN
    DuPont Pioneer



    I have worked with Kent on multiple projects. Those projects have been very successful because he has great attention for detail. He finds creative ways of either solving problems or optimizing existing solutions. These optimizations have helped reduce the necessity for large capital purchases.

    Adam Hock
    Unix SAN Manager
    Corteva Agriscience






    Kent A. Vander Velden, PhD
    kent@confluencerd.com
    70/30 Engineer-Scientist

    Kent has over 20 years of industrial and academic experience applying technology to improve the work flow of researchers. Before Confluence R&D, Kent led a team of informaticists developing data analysis pipelines for a large biotech company. Witnessing the potential for improving people's working lives led Kent to develop software and hardware systems for small companies. Kent loves learning, making, and helping others. These loves are incredibly complementary and the result shows.


    Melinda S. Vander Velden, MSEE, JD
    melinda@confluencerd.com
    General Manager

    Originally a systems engineer in precision agriculture, Melinda then earned her JD and has years of experience prosecuting biotech, mechanical, and electrical patents and negotiating contracts. Melinda also owns an IP law firm developing IP strategies for others.






    Our other family businesses

    Vander Velden Law Firm
    Practicing intellectual property law
    KvvCreates
    An electromechanical vlog