Project Motivation, Needs, & Goals

My name is Rebecca Schmitt and I am a 4+1 student at Duke University. I am currently in my undergraduate senior spring, and I am taking graduate classes now as a part of the MS in Mechanical Engineering program. I have worked on air quality research at Duke as well as Duke Kunshan University. For this class, I have been working towards a quadrupedal robot capable of air-quality monitoring.  

Since studying biodiversity and biogeography in Australia my first-year summer at Duke, I developed an appreciation for biomimetics. I was amazed at how practical so many things in nature could be. Biomimetics is essentially utilizing design concepts developed by the natural world and applying these properties to man-made technology, processes, and products. 

One branch of biomimetics is terradynamics, in which biology, robotics, and physics are synthesized to create machines capable of land locomotion.The benefits of a moving robot capable of traversing complex terrain are endless: there are search & rescue applications, defense and public safety, home service and health care, both of which would enable people, and environmental monitoring.

I am most interested in environmental monitoring and I believe that air sensors would be beneficial for the following reasons:

Poor Global Air Quality  

 Following global worries of worsening air quality in highly industrial cities, companies have even developed specialized facemasks that aim to protect users from air pollution. Upon viewing the map below, it becomes apparent that these concerns are quite valid and that this is one instance of public health precautionary measures. The map below uses levels of ozone and particulate matter concentrations to generate an Air Quality Index (AQI) for a given area. 

One of the most reliable indicators of air quality is particulate matter concentration; higher particulate matter creates a lower quality of air and vice versa. Particulate matter are tiny liquid or solid droplets in the atmosphere that originate from vehicular exhausts and various forms of combustion; these particles can then also chemically react with other components in the air (NYS Department of Health, 2016). When particulate matter is about 2.5 microns in diameter, it is referred to as either fine particulate matter or PM2.5. Though high levels of particulate matter as a whole is concerning for an environment’s air, PM2.5 poses an even deeper threat. Increased PM2.5 has adverse effects for public health as these tiny particles can become lodged in the lungs and aggravate respiratory health (NYS Department of Health, 2016). Lung irritation, chronic bronchitis, increased asthmatic issues, and other lung inflammatory responses are linked to high PM2.5 levels (NYS Department of Health, 2016). Particularly vulnerable populations are children, pregnant women, the elderly, and joggers (due to increased breathing).  

With such critical effects, the Environmental Protection Agency (EPA) and the World Health Organization (WHO) have standardized safe levels of  PM2.5 exposure for both long-term and short-term. The EPA’s current standards are 35g/m3 for a 24-hour period and 12g/m3for a yearly average (NYS Department of Health 2016). The WHO’s standards are a bit more strict, with 25g/m3for a 24-hour period and 10g/m3 for a yearly average (WHO, 2018). In 2016, 91% of the world’s population lived in conditions that did not meet WHO guidelines (WHO, 2018). Highest PM2.5 concentrations are seen in major industrial cities. With statistics like this, air quality surveillance becomes a public safety measure. 

Lack of Monitoring 

As seen in the map above, there are many areas in which researchers have limited data, with sparsely spread sensors. Some places even have no data at all. Of course, for some regions, there are geographical barriers such as the Australian Outback, the Sahara Desert, or mountain ranges. Other areas are remote. However, there are places in South America and Africa that are both well-populated and have no notable geographical hindrances, but still lack the air-quality monitoring support. Likely due to economic barriers, certain countries are not able to survey their air-quality at all. Hence, a low-cost sensor would be the most ideal for these places.  

Even in places where there are plenty of sensors to nationally track the air quality, the average citizen does not have the ability to measure air pollution in their direct vicinity. These PM2.5 and ozone concentrations may vary, but the AQI gives a good approximation for a region. Even so, someone may want to know where in their home or office building the air quality is the worst, so that they may optimize placement of air filtration systems. In my previous coursework, I was involved with researching how air quality varies on two different campuses, Duke and DKU.  

The air quality in Kunshan, China is widely perceived as poor, yet quantitative data is limited. In March 2018, Duke University undergraduates enrolled in CEE 292: International Applications of Environmental Field Methods traveled to Duke Kunshan University (DKU) to address this knowledge gap. The team utilized PM2.5 sensors in both Kunshan and Durham to compare indoor and outdoor PM2.5 concentrations and the effects of these levels on human health based on the biomarker of exhaled nitric oxide.

The experiment found that DKU’s air quality had PM2.5 concentrations about ten times higher than Duke’s in both indoor and outdoor locations. I’m sure that administration was aware of the region’s poor air quality, but these numbers gave them a more individualized record. Even for a private university, without the access to this type of technology, measuring the air pollution of a certain area is nearly impossible and one can only rely on the sensors already installed elsewhere for data.  

Environmental Justice 

Already briefly touched upon earlier but making this type of technology accessible and affordable would close gaps in knowledge. Besides the lack of environmental monitoring in certain countries, within specific nations it is often that only the most well-resourced areas will have sensors tracking the air quality. Typically, the more industrialized areas tend to have lower air qualitywithin these cities, it is often that marginalized people are more likely to suffer from prolonged exposure. PM2.5 is particularly harmful and has shown to cause asthma to those with prolonged exposure. It is an act of environmental justice to better equip vulnerable populations with the tools to keep track of and research their own living conditions. With this information, self-advocation and community rallying is much easier as there is data to support the need for policy-change.  

Having a movable air sensor would be highly beneficial. There is potential to create mappings of gas concentrations and even monitor areas that were too difficult/too expensive before. If a particular person is curious about the air quality in multiple areas, but cannot spend resources on multiple sensors, a moving air sensor would give them the ability to monitor many areas at different times without needing to physically move them themselves. PM2.5 and Ozone are dangerous for human health. Concentration monitoring can help people make informed decisions on minimizing their exposure as well as equip them with the technology needed to promote policy-change.  

Robotic Dog (Git Hub)

The larger goal of my project would be for the dog to be used as a research tool.  I would like to make a robotic dog that is able to move around and detect gas concentrations, but there are a variety of other data sampling applications. With added autonomy, this dog would be able to traverse areas dangerous for humans and collect data to send back to researchers. Other more advanced features would be implement SLAM (simultaneous localization mapping), identify and classify objects, or machine learning. The work for this semester serves as a foundation for these future goals. I was able to accomplish the following: assemble mechanical portions of dog, install software, control motors, and detect gas concentrations through raspberry pi.

Learning Objectives

 

The objective of this project is to build a robotic dog capable of walking and conducting data sampling. Through this project, someone will gain the following experiences: 
  • Basic electronics and 3D printing skills
  • Assembly of electronic and mechanical parts
  • Calibrate gas sensors, sample, and plot concentrations 
  • Motor control and calibration
  • Path-planning using Bezier Curves

Project Vision and Narrative

This project may be divided into learning modules of increasing difficulty. These lessons are not independent, but actually build on one another. The completion of one stage would be needed to continue. The first module aims to gain skills on 3D printing, mechanical assembly, and using gas sensors. The second module focuses on the electronics portion and the advanced module follows up on that with motor control. Finally, the expert lesson revolves around motor control and path planning for the robot’s gait. Credit goes to the creators of this GitHub, which was used as this project’s inspiration and main guide. 

Bill of Materials

This is a rather intricate project, so the bill of materials is lengthy. Additionally, there is a variety of tools that one must have access to in order to properly build the dog. In addition to the base of the dog, there are additional components needed for the gas sensing portion. Thus, I separated the list in case another application is desired bu someone wishes to build the dog on its own. To provide context, the original goal was for the dog to emit a barking sound if the gas concentration was above a certain threshold. To complete this project, you will need the following supplies and tools: 

Instructional Lecture

In this video, one will learn about Bezier Curves. First, the topic is introduced with useful applications. Then, the mathematical formulas are derived for the linear, quadratic, and cubic functions. Higher order Bezier Curves are shown without equations, but can easily be derived using the following formula: 

As explained in the video, the Bezier curve is calculated by a linear combination of smaller order Bezier Curves. These functions are all defined by Binomial theory.  After the math, the specific application for this project is shown as the original GitHub uses a 12-point Bezier curve. Finally, the motor control for my project is shown to further assert that path planning is essential for a successful gait. 

About the Author

I am currently an undergraduate senior double majoring in mechanical engineering and mathematics. I am in the 4+1 program, so I will be returning to Duke in Fall 2021 to complete my master’s of science in mechanical engineering. I grew up in South Florida and moved to Atlanta, Georgia during highschool. This is where my interest in environmentalism first began at Zoo Atlanta. I am primarily interested in designing and implementing ways to make renewable energy widely accessible. On campus, I have been apart of the Duke Smart Home as a resident and the president. I participated in Solar Spring Break with Grid Alternatives as a volunteer to install solar panels in residential homes.

I’ve had a variety of research experiences, but my most valuable one would have been with my Air Quality research team. Some of the research detailed here come from that class. On the other hand, this specific class has given me the opportunity to gain a more hands-on engineering experience. It was great to build this robot. It was definitely out of my comfort zone and unrelated to anything that I’ve done before. Besides figuring out how to fit this project into my own research interests, the project on its own was a fulfilling experience!