STEM Forum

Robots have evolved into our daily world from restaurants to manufacturing parts for planes. The evolution of robots is fascinating as researchers and engineers begin to build humanoids who can talk or rescue robots made from hydraulic fluid conducted in Boston Dynamics to even robotaxis! However, robots need specific code inputted into their hub, and this code has evolved into a new era of Machine Learning. What is Machine Learning? Machine Learning capabilities in the real world. Machine learning is like human learning where a model running on machine learning takes data and algorithms and trains on those models to get a better model that can now provide accurate predictions on future occurrences. Machine Learning is becoming the new era of code as Generative AI allows for the creation of content, design, code, and even influence business decisions. However, machine learning is not that simple. Libraries of ML Pandas: helps coders work with data and analyze and manipulate that data and allows for the simplification of data into a streamlined code that makes users to easily read, especially for a large training set. Keras: Simplifies the development of neural networks and allows for rapid prototyping on Computer Processing Units (CPUs) and Graphic Processing Units (GPUs). This is incredibly important as the use of computer chips becomes prevalent in ML models corresponding with its speed. Tensorflow: A major library used in almost all ML models that allows for the analysis of data and allows with the compatibility of Tensor Processing Units (TPUs) (Google’s circuits to boost ML workloads) and GPUs. One application of Tensorflow is in Tesla car’s when self-driving mode is on! Keras and Tensorflow are even used together as they do similar functions where Tensorflow allows for a larger scalability and boosts efficiency and processing. For example, when scaling data processing, you can use tf.data and Keras pre-processing layers to create a data pipeline. There are also Application Programming Interfaces (APIs) that allow different software to interact together, and APIs used in Keras or Tensorflow such as Tensorflow and Keras Sequential API to train neural networks. Scikit-Learn: A library of tools for machine learning which is generally used in Matpotlib and Numpy that allows for predicting decisions, gather data points together, identifying objects based on a set training model, evaluating models, and many more functions regarding data and ML model functions. Matpotlib: A library that allows for data to be viewed such as graphs, charts, and other visualizations of data. Scipy A library that allows for regression models to be inputted as well as math to be integrated into Machine Learning models. Numpy A library of Machine Learning libraries designated to manage large amounts of data. OpenCV A library used for computer vision. For example, cameras could use when determining if a person has littered on the road. How to make an ML model? A simplified example of an ML model using Scikit-learn Data Preparation Collecting data is crucial as that is what your model will be trained on in order to give you accurate predictions. Make sure to contextualize the data, whether it is images, texts, tabular data, etc. Clean data by filling in missing data, removing duplicates, and missing values. Finally, splitting data into training and testing sets is important because the training data is what the ML model is trained on to learn about predictions, patterns, and features from the data while testing data is used to evaluate the model’s performance by giving this unseen data and assess the model’s ability for its given function. Select a Model Choosing an algorithm for your ML model depending on your wanted function such as linear regression, neural networks, and decision trees. Neural networks are used in robotics as it is an alogirithm that suppossed to work similarly like the human brain. Functions such as computer visualization and prediction can be done through neural networks. Linear Regression is used to find a line that best fits with the given data used to predict continous values. Decision trees are used to make decisions; however, there are levels of a decisions given such as determining the outcomes, risks, resource costs, probability, etc. Train the Model Fit the model by using the training data to train the model and customize batch size and the amount of data wanted to be used (epoches) Overfitting and underfitting are problems that can occur during this process due to algorithms resembling the training data too closely or not enough respectively that it cannot give accurate predictions to the testing data. Assessing the Model Using the testing data in order to evaluate the model’s accuracy. Metrics: Assess accuracy, precision, recall, F1 score, mean squared error, or other relevant metrics, depending on the model’s purpose Modifying the Model Changing algorithms, training/testing data as new data comes out, cross-validating to improve the model’s performance, etc. Deploying the Model Store the trained model for future use and integrate the model in an environment where it is constantly doing its function on new testing data Different Environments include Jupyter notebook, Google Colab, and local IDEs What are Robots? Robots are machines that senses, processes, and actuates that interact with the environment. Sensing Sensing are typically done by sensors commonly done by cameras to visualize their surrounding environment, sound sensors that uses sound waves to detect obstacles around them, and other sensors such as pressure, temperature, etc. Processing Processing is the next step after sensors where the robot would then convert its input from the sensors into code that would then go to the control hub to be processed. Actuating Finally, actuating is the output after the processing is completed where it interacts with the environment and performs tasks whether that means to stop, drive, etc. Why are Robots Important? Robots have rapidly advancing in our new world of AI and changing our world before we know it. They are being used in factories to automate the build of cars, plane parts, and other manufactured parts. Furthermore, robots have been improving the way cars move, watering plants, and even serving food at restaurants. Its ability to conduct human tasks allows it to be constantly efficient and cheap. How are Robots Used with ML Models? Robots have started to become much more complex using ML models from being able to map out roads to identifying objects when functioning. Machine learning code often needs to be optimized for performance, real-time decision-making, and hardware constraints. Example Use Case: For example, when robots use cameras, robots have to navigate a world of obstacles among houses, factories, etc. A possible ML model to integrate into a robotics application would be using Python with libraries and TensorFlow for the model with a ROS (Robot operation System) as the robotic framework for object detection. Packages Installation Installing the necessary packages Loading a pre-trained model We imported Numpy as it can handle large amounts of data (needed when determining objects to give accurate predictions to the robot) and cv2 which is used for computer vision. TensorFlow is used for the analysis of the data inputted from the cameras themselves and build on the new data. After loading the pre-trained model (already trained with the training and testing data sets) we would define the function which is to detect the objects from the input of the camera. Here we are integrating the ROS for real-time operation by first importing the necessary software such as the CV bridge which converts OpenCV images to ROS messages while sensor_msgs.msg allows for common message definitions for robot sensors that help facillitate the communication of sensor data between nodes and importing “Image” means that it is being used for representing image data from a camera while containing metadata of the object itself: height, width, encoding type (allows for algorithms to easily recognize the data), and the image of the object itself. Finally, importing rospy is the ROS environment itself. We can then see a defined node and connect the ROS to the camera and using the CV Bridge, in the “image_callback” it converts the ROS messages into Open CV image formats that leads to the detection of objects using cv_image and then finally processing and publishing the images. Here is the full code: import rospy from sensor_msgs.msg import Image from cv_bridge import CvBridge class ObjectDetectionNode: def init(self): self.bridge = CvBridge() self.model = tf.saved_model.load(‘ssd_mobilenet_v2_fpnlite/saved_model’) rospy.init_node(‘object_detection_node’, anonymous=True) rospy.Subscriber(‘/camera/rgb/image_raw’, Image, self.image_callback) self.detection_pub = rospy.Publisher(‘/detections’, Image, queue_size=10) def image_callback(self, msg): # Convert ROS image message to OpenCV formatcv_image = self.bridge.imgmsg_to_cv2(msg, “bgr8”) # Detect objects detections = self.detect_objects(cv_image) # Process and publish detections detection_image = self.visualize_detections(cv_image, detections) detection_msg = self.bridge.cv2_to_imgmsg(detection_image, “bgr8”) self.detection_pub.publish(detection_msg) In the first definition, the image data is being converted into a TensorFlow tensor, then by adding a new axis for creating a batch dimension as TensorFlow expects inputs in batches, and then finally passing the image tensor into the model in order to provide detection results like bounding boxes, class labels, and confidence scores. The visualization of detections portion refers to processing the detections and overlaying visual elements (like bounding boxes) over the image and returning the image with the visual modification and essentially puts the data into a human-understandable way. Finally, the last part allows for the model to keep on running as incoming data continues to be inputted. The ObjectDetectionNode handles the detection and visualization process while the rospy.spin continues the ROS node to process incoming messages is constantly inputted. The except function finally allows for any errors or exception when the ROS node is disrupted. Here is the code: def detect_objects(self, image): input_tensor = tf.convert_to_tensor(image) input_tensor = input_tensor[tf.newaxis, …] detections = self.model(input_tensor) return detections def visualize_detections(self, image, detections): # Visualization logic here (e.g., draw bounding boxes) for detection in detections: # Process detections and overlay on image pass return image if name == ‘__main__’: try: node = ObjectDetectionNode() rospy.spin() except rospy.ROSInterruptException: pass Conclusion In conclusion, robots are advancing with ML making it more complex by making predictions, classifying images, and automating by themselves, and it is definitely going to be a huge step in humanity. Sources: Geron, Aurélien. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems. 2nd ed., O’Reilly Media, 2019. Kaehler, Adrian, and Gary Bradski. Learning OpenCV 3: Computer Vision in C++ with the OpenCV Library. 2nd ed., O’Reilly Media, 2016. Quigley, Morgan, et al. Programming Robots with ROS: A Practical Introduction to the Robot Operating System. O’Reilly Media, 2015. Robotics Casual. “ROS Tutorial: How to use OpenCV in a Robot Pick and Place task for Computer Vision.” Robotics Casual, 2022, https://roboticscasual.com/ros-tutorial-how-to-use-opencv-in-a-robot-pick-and-place-task-for-computer-vision/. Accessed 12 Aug. 2024. ROS Wiki Contributors. “cv_bridge/Tutorials.” ROS Wiki, ROS, 1 Feb. 2011, http://wiki.ros.org/cv_bridge/Tutorials. Accessed 12 Aug. 2024.

What are the questions? You must be wondering from the title. How is Nanotech truly going to change Climate Change? What are the applications? Is it even possible? The Devastating Problem Climate Change as you know has been affecting people and habitats globally. 3.3 billion people are highly vulnerable by this climate crisis. Policies in place by the end of 2023 would likely see temperatures exceed 1.5C this century and reach around 3.2C by 2100. There are already huge ice glaciers melting in the North and South Pole as well as Antartica. This could result in floods and waves and essentially wipe out the whole human existence! Scientists have had a huge debate on whether it is too late or not, but everyone is trying everything that they can in order to fight climate change. Imagine a man in Chad, one of the most vulnerable countries in the world towards Climate Change, having with the sweltering heat just beating constantly on this individual. There is constant rise for hunger and people are fighting for food as Climate Change kills the crops that farmers are making and droughts are ocurring. The individual will have to walk miles to find water in order to sustain him and his family; otherwise, it will be too late. He could see people dropping like flies because of having no access to water and food, despite having enough money that could help them to survive. Climate Change is inevitable as many gas emissions are used today in society to fuel everything we need. In many parts of the world, humans and ecosystems will be unable to adapt to this amount of warming, and the losses and damages will “escalate with every increment” of global temperature rise. This is why it is so important to at least reduce if not mitigating climate change today. So, What’s the Revolutionary Solution? I am glad you asked. The solution is something so small you cannot even see it. Something so minusicle you would not even notice it. This “thing” is in normal technology today, and it is probably in the device that you are reading from. That’s right, nanotechnology. What is Nanotechnology? Nanotechnology is what it sounds like, technology at the nanoscale. Want a reference of how small a nanometer is? Imagine 100,000 nanometers on 1 hair strand. In 2012, our smallest chip could reach 10,000 nanometers, but today, we can now reach to 1 nanometer for a single chip. Imagine 100,000 chips on a single strand of hair. This size and ability to store data and work with technology allows nanotechnology to be one of the most revolutionary industries in the world. Nanotechnology will basically affect every industry from filming to cybersecurity to medical fields. However, there are different types of nanotechnology designated for each one like subsections. The one used for Climate Change is nanoparticles. The Little Absorbers Nanoparticles contain larger surface area per unit volume, which adds an advantage with transporting clean energy and adsorbing greenhouse gases. This higher surface area allows for more CO2 emissions to be absorbed as emissions have more interactions with these nanoparticles. Nanoparticles have special properties that allows them to regenerate. This regenerative property allows to them to last longer into absorbing the emissions that are so prevalent in engines and in the general environment. The Verifying Experiment Nanoparticles of Fe3O4, Fe3O4–proline, Fe3O4–lysine, and Fe3O4@SiO2–NH2 were placed in water. The effects of initial pressure of CO2, nanoparticles and nanoparticle concentration on absorption performance were examined. The obtained results indicated that the presence of Fe3O4 nanoparticles resulted in better results at pressure 30 bar in water as a physical absorbent, and they enhanced the absorption up to 17.61% and 34.23% compared to non-functional Fe3O4 water, respectively. Furthermore, the dispersed nanoparticles in the MDEA nanofluid shows that the functional Fe3O4 nanoparticles were effective for CO2 separation even in chemical absorbents, and under the best conditions, they enhanced the capacity of absorption up to 16.36% at pressure 40 bar. Moreover, zeta potential analysis, the graphs used into seeing if the nanoparticles provide stability in the base fluids, revealed that Fe3O4–lysine and Fe3O4@SiO2–NH2 nanoparticles are more stable than non-functional Fe3O4 and Fe3O4–proline in the base fluid. In this picture, this resembles the summary of the experiment that was carried out by these magnetic nanoparticles. Bubble breaking as shown by the picture occurs when intially, it start with the separation of CO2 and nanoparticles, but eventually, the nanoparticles start to become unstable, thus testing the stability of nanoparticles at base fluids using the zeta potential analysis in order to cause the reaction for the carbon emissions to break. Another method as shown by the picture is the Brownian Method. The Brownian method is essentially a sporadic movement of particles suspended in a solution due to the interaction between the particles and fast-moving atoms in gas or liquid. As shown in the diagram, the carbon is clustered in the middle or in the medium where it is trapped in the middle and preventing the carbon emissions from escaping. The last method shown in the picture is the Grazing Effect. The Grazing Effect is where nanoparticles diffuse across to the carbon, and it is absorbed directly to on the surfaces of the nanoparticles. This allows for the carbon to be trapped within the nanoparticles as it moves across the gradients. These methods were tested during the experiment as well when figuring out the ways these magnetic nanoparticles of Fe3O4, Fe3O4–proline, Fe3O4–lysine, and Fe3O4@SiO2–NH2 could prove these methods of absorbing carbon emissions. Nanoparticles Teaming with the Ocean The ocean is so vast, reaching up to 139 million square miles, and one of its main functions is reducing the amount of climate change in the environment. Not only does the ocean cool down Earth and remove the heat from the environment, but they also reduce the carbon emissions present using these small living organisms called phytoplankton. Phytoplankton converts carbon emissions through photosynthesis into living tissue, an amazing way to turn something deadly into something reviving. The phytoplankton then sinks to the sea floor, taking the absorbed carbon into the depths of the ocean for centuries to come. But how does nanoparticles come into play? Glad you ask. So, phytoplankton requires a fertiliser or combination of iron and other nutrients to be given in order to flourish, and scientists have faced the problem of providing these nutrients. One way scientists have tried was directly pumping these nutrients into the ocean, but this proved to be inefficent or if it did help the phytoplanktons to flourish, it would not absorb as much carbon dioxide as natural blooms occured. A new scientific paper authored by Dr. Peyman Babakhani from the University of Leeds debates that these problems can be solved if engineered nanoparticles to deliver fertiliser to the phytoplankton and would be cost-beneficial as they have the ability to regenerate and would not waste an excess amount of resources. Directly quoted by the paper, “Our study [, based on 123 studies previously,] shows that the use of several types of engineered nanoparticles in ocean fertilisation can be promising in terms of costs and carbon dioxide emissions during production and delivery processes, and such nanoparticles can be applied at concentrations lower than those that might cause toxicity to marine ecosystems.” As you have read from the last sentence of the quote, there is a public fear of the nanoparticles being toxic to the marine environment as they are engineered particles being placed into a natural ecosystem and requires assessments before putting them into the ocean. However, scientists believe that turning nanoparticles into become non-toxic is possible to achieve. The ocean has been a big asset into fighting climate change from the very beginning, and sceintists has been researching for years how we could further enhance our ways into making the ocean a powerhouse absorber. They wanted to find another way to make phytoplanktons, the main reason why the ocean absorbs 25% of carbon emissions on Earth, but this could even turn to double if not triple of carbon emissions being absorbed once we have harnessed this new technology that is being implement by marine specialists and nanotechnology scientists. The Bubble Breakup Among many of the methods of reducing carbon emissions such as absorption or membrane separation, the most popular and revolutionizing is using organic liquids to absorb CO2 from gas as it proves a huge decarbonization and a simple process to be implemented. In particular, the addition of the nanoparticles in the gas-liquid flow is a viable way in removing carbon emissions by improving the mass transfer across the gas-liquid interface. The high CO2 absorption rate can be obtained under room temperature by adding Al2O3 nanoparticles into the methanol absorbent specifically. Bubble columns provide a large interfacial area that would be used for the nanoparticles in order for them to interact with the carbon dioxide. The gas is consists of bubbles via the holes at the bottom of the column that then travels through the liquid. Bubble coalesence and breakup are interaction that affect the performance of contactors (systems that bring different phases, generally gas and liquid, for mass transfer or reactions). This determines the bubble size and eventually the bubble rise velocity by the change of the interfacial area and gas holdup, leading into. influencing the interphase mass transfer rate. Challenges include consistently keeping efficient flow conditions, requiring an understanding of bubble interaction dynamics and the mass transfer interactions The first basic experiment involving bubble coalescence was conducted by Calderbank et al. It was agreed that coalescence occurs when the following bubbles gather behind the leading bubble, where the bubble wake plays a vital role in both capturing non-aligned bubbles and the possible subsequent coalescence In recent decades, computational investigations have been implemented to experiment the changes in geometric structure over locations and time and through the interactions among and between discrete phases and fluid flow. In particular, the population balance model (PBM) can predict the local size distribution of fluid entities by accounting for the coalescence and breakup. The application of the PBM is significant to predict the bubble population dynamics gas-liquid systems Now, how does this correlate with nanotechnology? An experiment was conducted in Korea University by Yong Tae Kang and Lirong Li consisting of, the bubble coalescences and breakup interactions in a rectangular bubble column for CO2 absorption in methanol with various concentrations of Al2O3 nanoparticles, creating the gas-liquid system. A high-speed camera and simultaneously visualized via simulations with the PBM to determine bubble size distributions. Then, a series of parameters correlated with the bubble interaction behaviors, such as the bubble velocities, bubble wake patterns, path instability, and frequencies of bubble coalescence and breakup, were examined. The mass transfer coefficient and the bubble behaviors is also considered for the carbon dioxide to be transferred to these bubble interaction in order to be absorbed by these nanoparticles. In conclusion, the nanoparticles proved to be incredibly effecient for the carbon dioxide absorption based on the gas-liquid nanoparticle model system that was utilized during the experiment, and this was extremely important experiment as it proved the way fluid dynamics could be essential with the use of nanoparticles in order to eradicate carbon dioxide. Final Conclusive Thoughts As we have seen on how nanoparticles are essential to the defeat of climate change, it is in the power of scientists and engineers to capitalize on this opportunity of nanotechnology to absorb carbon dioxide and further enhancing the way we already decarbonize the environment. Now do you believe me how something so small can have a tremendous impact?

20,000 unnecessary deaths from Lymphoma each year in America alone, 300,000 unnecessary death worldwide with no real and effective solution to prevent it from happening? It is crazy to think such drastic effects are caused by just one cancer. Is there a viable solution that can prevent such death from happening? Lymphoma So, what is Lymphoma? Well, lymphoma is a cancer that starts in the lymph system that have two main types: Hodgkin Lymphoma and non-Hodgkin lymphoma. Non-Hodgkin lymphoma is a an unpredictable cancer that can appear in any lymph nodes in the body while Hodgkin Lymphoma is more predictable and happens in the upper body like the neck and chest. Lymphoma is a horrible cancer because it makes us more susceptible to infection and diseases such as heart disease as lymphocytes or lymph nodes control immune response or produce killer T-Cells. Cancerous lymph cells multiply uncontrollably just like any other tumor, but in the case of lymphocytes this creates an inability for the body to fight infections with healthy T-cells — so the treatment process becomes counterintuitive since chemotherapy is meant to reduce this multiplication effect, but if the body isnt able to adjust to unhealthy lymph tissue, its a lose lose situation PDTs Photodynamic therapy (PDT) is a method used to eradicate cancerous cells by using drugs that is activiated by light as a or a laser such as LEDs are known as photosensitizers to kill cancerous cells. Cells would essentially have absorbed photosensitizers that would be set to a specific wavelength of light which would then produce “oxygen radical” that would begin to kill off the cells. The drug would injested from the mouth, spread on the skin, or given through an IV, depending on where the tumor is in the body. Most of the drugs would leave normal cells after 1–3 days but remain in pre-cancer or cancerous cells which would then be exposed to the photosensitizer. Depending what type of cancer it is will be shown to direct the light like for skin cancer, the light would be directed towards the direct center. It would then create that form of oxygen that kills cells Huge Drawbacks: Photosensitizers in photodynamic therapy cannot pass through more than 1/3-inch of tissue. So, photodynamic therapy can only be used to treat tumors that are on or “just under the skin or on the lining of internal organs or cavities” (National Cancer Insititute). Furthermore, many side effects occur from PDT such as burns, pain, swelling are all possible even when PDT is very limited. Other side effects include: • Painful breathing • Shortness of breath • Eye Irritation and redness • Skin irritation and becoming more susceptible. Benefits: It is much more precise than many other cancer method and prevents much killing of healthy cells like chemotherpay does. However, it not reaching the needed areas ( more effective with precision but not so much as being effective to kill the tumor itself unless small). What are Researchers doing Today? Newer delivery of photosensitizers that allows for more precision with an immune response that directs the immune system into eradicating the tumor known as photoimmunotherapy or PIT. An immune protein is combined with a photosensitizer Researchers are also developing a new type of PDT called photoimmunotherapy, or PIT. In this treatment, a photosensitizer is combined with an immune protein that delivers the photosynthesizer to cancer cells. When light is applied, the photosynthesizer kills the cancer cells. This process also causes an immune response inside the tumor that can cause more cancer cells to die. “ This new immune response allows for deeper and larger tumors to be accessed while being even ore precise and more powerful against those tumors However, in Lymphoma cancer, there is already a weak immune response to begin with, preventing any photoimmunetherapy to effectively get rid of the virus, rendering this amazing solution into uselessness. New Solution: UCNPs Up-conversion nanoparticles (UCNP) has been a new recent development where these nanoparticles are able to used near-infrared (NIR) light that activates UCNPs to emit high-energy light (commonly UV light) that activates the surrounding photosensitizers. These photosensitizers would then make singlet oxygen and other reactant oxygen species (ROS) which would kill of the designated areas that is involved. This NIR with UCNPs allows for scientists to be able to lessen the side effects and become more effective by being able to penetrate up to 3x more than a regular UV light that would be used in regular PDT treatment as well as lower the phototoxcity that was inflicted in nearby normal cells and tissue that other method like chemotherapy would have infected. How would this all work? UCNPs as energy donors will have to first be loaded onto the photosensitizer (PS) molecules for light having energy from donor to acceptor in NIR- induced PDT. Silica Encapsulation is one of the most effective methods that are used to load UCNPs on PS molecules by coating UCNPs with a silica shell. For example, a group of researchers had developed NAYF4 UCNP carriers for zinc(II)-phtalocyanine (ZnPc). Through physical absorption, the ZnPc was encapsulated after the UCNP was encapsulated, creating a stable and airtight encapsulation. Finally, under a 980-nm NIR light, nanoparticles were excited to create singlet oxygen that would destroy the cancer cells. After loaded with the ZnPc molecules, the 980-nm NIR light would then activate the ZnPc molecules through an energy transfer from the UCNPs as it would emit the visible light needed because of the NIR light. This posed a problem as an experiment with this was conducted, but it created singlet oxygen that would attack DNA bases and mitochondria in our cells during the 5 minute NIR irraidation time and the laser going at 0.5 W. However, Shan et al. have done a study that had increased the ratio of photosensitizers that were loaded on the UCNPs that had made up 10% of the total weight while encapsulating 100nm of UCNPs and using meso-tetraphenyl porphine photosensitizers that allowed much DNA and mitochondria from being killed. Through this newly set up experiment, Shan, a photodynamic researcher, along with many of his peers, had used “980-nm NIR light excitation at 134 W/cm2 for 45 min,” effectively killing the cancerous cells (Shan et al.). Using these new photosensitizers and varying the ratio size, Shan was able to avoid creating such dangerous singlet oxygen that ZnPc was producing and lessening the amount of PS molecules compared to the UCNPs. These researchers concluded that NIR excitation with a UCNP-PS nanocomplex offered a incredibly productive way to kill off cancerous cells. To further enhance tissue penetration, a Ce6 solution was created in recent nanomedicine studies that would combine with UCNPs. Using pork tissue to test it’s effectiveness, a UCNP with a Ce6 sample was ineffective to penetrate even 3mm of pork tissue; however, with the addition of the 980-nm light, it created significant amount of oxygen production even when it was blocked with an 8mm pork tissue. However, even though 980-nm NIR light is the most effective excitation as it offers effective tissue penetration to the extent of 3.2cm, it causes skin heating that could damage nearby tissues as water absorption’s peak is 980- nm. 915-nm NIR light acts just like 980-nm NIR light and offers a deeper tissue penetration. (Zhan et al). and avoid such skin heating to occur in UCNP-based imaging and PDT treatment. While UCNPs work in PDT, many imaging software are used in order to follow these particles and image tumors such as UCL imaging, so not only does UCNPs have the potential of eradicating tumors but also offers the opportunity to at least image the tumor to determine the extent of the tumor. How would the UCNPs be Delivered? This nanobot has three wheels, two lined with motors (shown with the metal plates) while the third wheel (the one across from the coil) acts as a stabilizer wheel to prevent the machine to tip over. The top dome to where this structure meets the cylinder (signified by the line) displays the payload, carrying the necessary UNCPs that have the photosensitizers to eradicate the tumor. As you can see, there is a hole on the bottom of this machine. With a pipe of metal running from the payload to this hole, these UCNPs will be able to be directly deployed on these tumors caused by Lymphoma anywhere in the body. The payload will be equipped with an automated seal in order to prevent any UCNPs from getting out of the pipe too early from the cancerous site. Finally, the coil will be used as self-propelled particles called “nanoswimmers” discovered by researchers from the University of Colorado Boulder. These coiled structures have already been tests by these researchers that prove its effectiveness in the overall environment of the body. This nanocarrier can totally revolutionize the deployment of UCNPs to specific tumor sites and allows Is it Possible in the Future? The only concern about this solution is the nanotoxcity that can occur in our bodies; however, many scientists are developing new ways with biocompatible coatings that would be used in UCNPs, and with the recent development of nanomaterials, researchers are confident to be able to develop such substances in just a few years! Our solution is easy to do as it has already been tested and trialed before while achieving the ultimate effectiveness that is needed in order to offer a deep penetration of tissue while limiting the pernicious effects of these methods. Ce6 solution, NIR light set to 915-nm NIR, meso-tetraphenyl porphine photosensitizers, and UCNPs with silica encapsulation are all components in our model to create a solution to Lymphoma. Works Cited: National Library of Medicine. (2013). Upconversion nanoparticles for Photodynamic therapy and other cancer therapeutics. PubMed Central (PMC). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3645058/#B44 Shan JN, Budijono SJ, Hu GH, Yao N, Kang YB, Ju YG. et al. Pegylated Composite Nanoparticles Containing Upconverting Phosphors and meso-Tetraphenyl porphine (TPP) for Photodynamic Therapy. Adv Funct Mater. 2011;21:2488–95. Zhan QQ, Qian J, Liang HJ, Somesfalean G, Wang D, He SL. et al. Using 915 nm Laser Excited Tm3+/Er3+/Ho3+-Doped NaYbF4 Upconversion Nanoparticles for in Vitro and Deeper in Vivo Bioimaging without Overheating Irradiation. Acs Nano. 2011;5:3744–57. You Exec. “The future of “nanobots” in 2023 and beyond.” YouTube, 2023, www.youtube.com/watch?v=oFH3gFFV0AI&t=462s. Accessed 21 Apr. 2024.