EnviTronics Lab

  Introduction When deploying Raspberry Pi cameras for long-term, unattended field applications, power management becomes a critical consideration. These devices can consume significant power, especially when coupled with additional modules and wireless communication. This blog post will explore the challenges of powering Raspberry Pi cameras in field conditions and present a solution for optimizing power consumption and enabling continuous operation. Power Consumption Challenges Raspberry Pi 3 models, even at low CPU speeds, draw a substantial amount of current. Connecting additional modules and establishing wireless communication can further increase power consumption. Preliminary measurements revealed that a typical field imager, based on a Raspberry Pi, could draw over 500 mA of current. This level of consumption quickly depletes a 7000-mAh acid battery, even with an efficiency of 85%. Power Management Solution To...

  Remember how a few years back we discussed a leaf wetness sensor (like the  INSHU LWS ) and the potential for a similar sensor for fruits? Well, today I'm excited to share a design concept I've been working on – a  DIY fruit wetness and temperature sensor  specifically for small fruits like blueberries!  Why a Fruit Wetness Sensor? Fruit surface wetness plays a crucial role in various aspects of fruit health. Excessive wetness can lead to cracking, disease development, and ultimately, reduced yield and quality. Imagine having a sensor that mimics a small fruit, allowing you to precisely measure wetness levels and prevent these issues. The Design Concept  This sensor design utilizes a two-pronged approach (Fig. 1):  Mimicking a Fruit:  The core is a 3D-printed "filler" shaped like the top and bottom halves of a fruit. PCB Electrodes:  A two-layer printed circuit board (PCB) with strategically placed electrodes (think fruit slices) is embedde...

  I'm a strong advocate for plants as the best soil moisture and potential sensors . Data-driven tools like this one integrating canopy feedback provide more accurate and reliable information for crop management compared to traditional soil-based techniques. Raw data alone isn't enough. Predictive modeling that interprets this data, along with feature engineering and data cleaning techniques, is essential. Biophysical and statistical modeling approaches both have their merits, but a combined strategy offers a more complete picture.  Plant monitoring through proximal canopy sensing is surprisingly less widespread than soil-based methods. Cost isn't the barrier - proximal sensing can be just as affordable as soil sensors. Satellite and drone imagery, while valuable, lack the necessary temporal resolution. We need 24/7 monitoring, similar to soil sensors, that proximal sensing offers. Perhaps the biggest obstacle to wider adoption of plant-based methods is the lack of cohe...

  I'm excited to announce the availability of a Python version of the CWSI and transpiration module/class originally developed in C++ for embedded systems (see this post: 𝗕𝗶𝗼𝗽𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗠𝗼𝗱𝗲𝗹𝘀 𝗳𝗼𝗿 𝗖𝗪𝗦𝗜 & 𝗧𝗿𝗮𝗻𝘀𝗽𝗶𝗿𝗮𝘁𝗶𝗼𝗻 𝗖𝗮𝗹𝗰𝘂𝗹𝗮𝘁𝗶𝗼𝗻𝘀 (𝗖++) ). 𝗪𝗵𝗮𝘁 𝗶𝘀 𝗖𝗪𝗦𝗜? CWSI is a reliable indicator of plant water stress, offering a valuable alternative or complement to traditional soil moisture and soil water potential sensors. 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 𝗧𝗿𝗮𝗻𝘀𝗽𝗶𝗿𝗮𝘁𝗶𝗼𝗻  𝗔𝗰𝘁𝘂𝗮𝗹 𝗧𝗿𝗮𝗻𝘀𝗽𝗶𝗿𝗮𝘁𝗶𝗼𝗻 (𝗧𝗮): The amount of water a plant loses through its leaves to the surrounding air.  𝗣𝗼𝘁𝗲𝗻𝘁𝗶𝗮𝗹 𝗧𝗿𝗮𝗻𝘀𝗽𝗶𝗿𝗮𝘁𝗶𝗼𝗻 (𝗧𝗽): The maximum amount of water a plant could lose under ideal conditions. 𝗕𝗶𝗼𝗽𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵 By taking a biophysical approach, these models are designed to be site-independent, making them potentially applicable to a wide range of crops. However, crop-spec...

  I recently read an article suggesting that OpenAI's o1 model could significantly impact the future of traditional coding. Interestingly, Jensen Huang, the CEO of NVIDIA, has expressed a similar sentiment, predicting that the rise of AI might lead to a decline in coding jobs. He even suggested that young people consider exploring alternative career paths such as biology ,  education ,  manufacturing , or agriculture . It's ironic to consider that machine learning, a field designed to automate tasks, including aspects of coding, is now being perceived as a threat to the very profession it was intended to assist. This old post ( How Artificial Neural Networks Are Convoluting Agricultural Engineering ) still relevant, isn't it? The post concluded that the application of AI in agriculture often prioritizes academic and professional advancement over addressing real-world challenges. This trend leads to several negative co...