Progress 03/15/17 to 03/14/21
Outputs
Target Audience:Graduate assistantship: Three M.S. level and one Ph.D. level students participated into this project. M.S. students successfully completed their thesis defenses in December 2018 and in March 2019. The students were involved in protocol development, determining inoculation methods, experimental design, conducting experiments, data collections and processing, training convolutional neural networks. Extension/Outreach: The research project has been mentioned in the university's annual research report and gained attention of many students. 2 EE, 1 ME and 1 CS students learned how to build these robots. We have also stated designing the course outline for the Objective-4. (Redesign of existing COMP/ENGR 4440 Mobile Robotics course). Also, our lab (Bioinformatics and Agricultural Robotics Lab, https://barlab.academy) is the most visited lab in the university and always our group has been given the task to do the lab tours to showcase the quality of research is being done in our university. (Some sample visits: Commanding General MG Cedric Wins from the U.S. Army Research, Development and Engineering Command, Division of Research and Institutional Advancement Nashville Chamber of Commerce Research and Innovation Tour and many high and middle school students) Changes/Problems:We were planning to use it in a real water pipeline but due to the pandemic we had to test the system in our lab environment. We have successfully improved the ego-location with fusing IMU data and image processing, but we were unable to test this approach in both the lab environment and in the real-life scenarios. What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?We have shared the experiment data and the code in our website with the public. The link is:https://barlab.academy/index.php/products/leak-detection-robot-deep-learning-based-motion-segment-estimation. Also, our lab (Bioinformatics and Agricultural Robotics Lab, https://barlab.academy) is the most visited lab in the university and always our group has always been given the task to conduct lab tours to show the quality of research is being done at Tennessee State University. (Some sample visits: Commanding General MG Cedric Wins from the U.S. Army Research, Development and Engineering Command, Division of Research and Institutional Advancement Nashville Chamber of Commerce Research and Innovation Tour and many high and middle school students) What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
Research Goal: We were able to design and implement a prototype of the MICR, but unfortunately, we were unable to test it in a real pipeline due to the pandemic. We designed a simulated pipeline in our lab to test the device and published the results in the paper mentioned in the publications section. Objective-1: Design and implement a Meso-Scale Intelligent Capsule Robot. The aim of this study was to create a fast, accurate, and cost effective meso-scale intelligent capsule robot (MICR) to detect leakage in water pipelines. The proposed robot is designed as an aerodynamically stable, lightweight, and waterproof robot to travel in water pipelines easierand steadier. It consists of major sensors (microphones, digital camera, inertial measurement unit (IMU)) as well as microcontroller unit (MCU), lithium-ion rechargeable batteries, SD card, and 3D printed parts. The sensory information is stored in SD card for post-processing. The robot's orientation and position data are recorded with a mounted inertial measurement unit, which measures angular rate in degrees per second, acceleration, and changes in its magnetometers. The novel approach of this project is (1) to localize MICR robots to inspect leakages in water pipelines and (2) to couple sensory data (acoustic, image) with IMU to localize the leakage. 1.Localization via Inertial Measurement Unit: Inertial measurement unit consist of three individual measurement devices, accelerometer, gyroscope, magnetometer. All these devices generate 3-dimensional vector measurement in cartesian space indicating the measured variable magnitude and direction. These measurements create valuable information to investigate the robot motion characteristics and behavior. Also, it allows toroughly estimate the position of the robot inside the pipe with respect to time. The utilized approach is directly based on one of the most popular, modern data processing technique, deep learning. The main criteria on choosing an IMU is that the device utilized should be in low dimensional, with low energy consumption. BNO055 is one of the most popular, highly reachable, cheap IMU devices and supports. To construct a training dataset for a pipeline motion segment detection system, it is necessary to define the general pipe structure in which the leak detection robot will be travelling. To make a general assumption for the proposed idea, a pipeline system could be considered as constructed of finite geometric shapes. The structure simply constructed by straight segments and bending segments. From the travelling MICR's point of view, also from the core motion measurement device (IMU), the traveling path contains straight segments, right turning segments and left turning segments. Due to this finite state geometric constraint on the pipe structure, it is possible to separate measurements taken from the IMU device, which also represents the traveling path of the robot inside the pipe, into 3 different classes. This idea leads to generate a training data set containing these 3 classes, straight, left turn, right turn. The developed idea depends on the generation of artificial time graph images for the considered sensor measurement variables. The created time graphs (related to the measurement instance) are successfully fed into the trained convolutional neural network model. The classifier is responsible fordistinguishing each individual measurement's motion type, straight, left turn, right turn, by analyzing the segment which takes the interested measurement point as its center. This approach allows the MICR to obtain knowledge, within time frame information, about the traveling action through a pipeline system. 2.Leakage signals: Previously, hydrophones (i.e. microphone designed to be used underwater) were used forlistening to recording acoustic data (20-20000 Hz) within water. This analog data was converted to digital using commercial analog-to-digital converter sound card of computers. It was shown that the sound originated from leakages and cracks has low frequency (peak frequency is about 105 Hz) and there is almost no content above 5000 Hz. The upper frequency limit of sound signals coming from the cracks and leakage depends on the type of the pipe material, flow rate, and pressure. These outcomes were verified by utilizing the planned microphone array (ReSpeaker Mic Array) and obtained satisfactory results which indicates signal strength and phase variation between different sound recording channels. However, as indicated beforre,the important issue is to rebuild the microphone array within a tiny package. Objective-2: Design and implement a control algorithm for MIRC. The development and application of active locomotion for Meso-Scale Intelligent Capsule Robots have now reached a level of maturity where significantly enhanced control and inspection functionalities can be achieved for capsule robotics that were not available a decade ago and often thought impossible. This research focused on (1) a passive control and (2) active control of these robots inside the pipes. 1.Passive Control: The first prototype of the robot control was a passive control in which the robot was moving with the flow of the water inside the pipes. The first results of this system are shown in the previous sections. 2. Active Control: The objective is to design and optimize a flagella-propelled bio-inspired submersible robot that can operate in hydroponic systems to assess the pipe zone health and operate in irrigation pipes to detect leaks by data fusion and image processing techniques. This system can also be used for the health assessment of roots in hydroponic farms. The research team designed a flagella-propelled submersible inspection robot and verified its maneuverability and operability under/around/through plant roots. Objective-3: Design and implement of an intelligent data management system. The first prototype of the system was designed to find the motion trajectory of the leak detection robot which was a valuable information to estimate the localization of an autonomous robotic system, especially in a very dynamic but structurally unknown environments like water pipes where the sensor readings are not reliable. The focus of this approach was to estimate the location of meso-scale robots using a deep-learning-based motion trajectory segment detection system from recorded sensory measurements while the robot travels through a pipe system. The idea was based on the classification of the motion measurements, acquired by inertial measurement unit (IMU), by exploiting the deep learning approach. Proposed idea and utilized methodology are explained in the related sections and it is observed that convolutional neural network approach is quite powerful to overcome the unreliability of IMU data. The previous results of the system were reported in the paper, "A deep learning approach for motion segment estimation for pipe leak detection robot". The second prototype of the system, the ego-motion of the leak detection was improved by using image processing and fused the information. We have successfully combined IMU with image processing and improved the accuracy of the localization of the MICR. Objective-4: Redesign of existingCOMP/ENGR 4440 Mobile Robotics course We updated the COMP/ENGR 4440 Mobile Robotics course with the IMU data and localization techniques using the expertise, which we gained from this project. We were not able to offer this course, but we were able to discuss the development and use of pipe inspection robots and shared the data collected for the leak detection in the COMP 5800 Data Visualization class. Also, we have added a module to the COMP 3650 Microprocessors course about the localization of robots using IMU. 20 students benefited from this module and one of the student groups did a project related to this.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2019
Citation:
Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Ali Sekmen,
A deep learning approach for motion segment estimation for pipe leak detection robot,
Procedia Computer Science,
Volume 158,
2019,
Pages 37-44,
ISSN 1877-0509,
https://doi.org/10.1016/j.procs.2019.09.025.
(https://www.sciencedirect.com/science/article/pii/S1877050919311834)
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2019
Citation:
M. Christian et al., "Application of Deep Learning to IMU sensor motion," 2019 SoutheastCon, 2019, pp. 1-6, doi: 10.1109/SoutheastCon42311.2019.9020363.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2019
Citation:
N. T. Tekin, E. Kaplanoglu, E. Erdemir, F. Baysal-Gurel, C. Uyanik and S. K. Hargrove, "Modeling And Testing Of Magnetic Speed Controlled Submersible Robot For Hydroponic Production," 2019 SoutheastCon, 2019, pp. 1-4, doi: 10.1109/SoutheastCon42311.2019.9020356.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2018
Citation:
Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection.
Fulya Baysal-Gurel, Erdem Erdemir, Tim Darrah, Erkan Kaplanoglu, Robert Turner and S. Keith Hargrove, Design of a
flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection , ICPP 2018
|
Progress 03/15/18 to 03/14/19
Outputs
Target Audience:Graduate assistantship: Since Fall-2018, two M.S. level and one Ph.D. level students participated into this project. M.S. students successfully completed their thesis defenses in December 2018 and in March 2019. The Ph.D student is currently pursuing his courses while working on the project. The students involved in protocol development, determining inoculation methods, experimental design, conducting experiments, data collections and processing, training convolutional neural networks for the projects. Extension/Outreach: One of the scientists in our team presented a talk in Tennessee State University Conference Series event and took attention of many students. We have now 3 EE and 1 CS students have learned how to build these robots. Also, our lab is the most visited lab in the university and always our group has been given the task to do the lab tours to show the quality of research is being done in our university. (Some sample visits: U.S. Army Research, Division of Research and Institutional Advancement Nashville Chamber of Commerce Research and Innovation Tour and many high and middle school students) Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Three undergraduates have learned 3D Printing, one has learned PCB board design and soldering. We are still working with these students and designed mobile robotics course. Also, two graduate students have become familiar with sensory measurement processing analyzation through deep learning. How have the results been disseminated to communities of interest?We have shown the process results of our project in following publications Matthew Christian, Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Sambit Bhattacharya, Rashad Bailey, Kazuhiko Kawamura, S. Keith Hargrove. "Application of Deep Learning to IMU sensor motion", Southeastcon, 2019, IEEE N.Turgay Tekin, Erkan Kaplanoglu, Erdem Erdemir, Fulya Baysal-Gurel, Cihan Uyanik and S.Keith Hargrovey. "Modeling And Testing Of Magnetic Speed Controlled Submersible Robot For Hydroponic Production", Southeastcon, 2019, IEEE. Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Ali Sekmen. "A deep learning approach for motion segment estimation for pipe leak detection robot". 3rd World Conference on Technology, Innovation and Entrepreneurship, 2019, Elsevier Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Sambit Bhattacharya, Rashad Bailey, Matthew Christian, Kazuhiko Kawamura, S. Keith Hargrove. "Deep Learning Approach for Motion Trajectory Estimation for Robotic Systems". International Conference on Robot and Human Interactive Communication, 2019, IEEE Fulya Baysal-Gurel, Erdem Erdemir, Tim Darrah, Erkan Kaplanoglu, Robert Turner and S. Keith Hargrove, "Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection", ICPP 2018 What do you plan to do during the next reporting period to accomplish the goals?Objective 1): Designing and implementation of a Meso-Scale Intelligent Capsule Robot We have finished %80 percent of this objective and will finish the rest next year. Objective-2) Design and implement a control algorithm for MIRC We have finished %70 percent of this objective. we have now a working on to rebuild the controller in smaller package size to successfully insert into the robot. Objective-3): Design and implement of an intelligent data management system We have finished data management system architecture and most of the component of the overall system is developed individually. The remaining sections are related to the real time application data which is going to be acquired after the robot hardware completed. Objective-4): Redesign of existing COMP/ENGR 4440 Mobile Robotics course. It is mostly designed by the help of the undergraduate students who works with us in lab. It going to be finalized in next year.
Impacts
What was accomplished under these goals?
Objective-1) Designing and implementation of a Meso-Scale Intelligent Capsule Robot Developing a cost effective meso-scale intelligent capsule robot (MICR) to investigate a water pipeline and detect leakages is the main focus of this study. The developed robot is supposed to travel inside water pipelines. The robot requires to contains sensors to collect environmental information (e.g. microphones, digital camera) and to collect the performed motion information such as orientation and position through the pipe with a mounted inertial measurement unit, which measures angular rate in degrees per second, acceleration and changes in its magnetometers. An SD card is planned to store the measurement data for offline investigation (i.e., auto-interpretation, localization) by software algorithms. A microcontroller unit is required to be utilized as master controller of all these sensors and also storage unit. To energize all these devices lithium-ion rechargeable batteries must be utilized. The novel approach of this project is (1) to use MICR robots to inspect leakages in water pipelines and (2) to couple sensory data (acoustic, image) with IMU to localize the leakage. The most important issue in the project all of these sensory devices, controller unit, batteries are required to be resigned to construct them into a smaller form. Inertial measurement unit and camera has already found in smaller form., in smaller form, the accuracy and trustability of these devices decreases dramatically. Some of the sensors such as microphone array, SD card adaptor, and microcontroller device is still in progress to rebuild into a tiny package. 1.Localization via Inertial Measurement Unit Inertial measurement unit consist of three individual measurement devices, accelerometer, gyroscope, magnetometer. All these devices generate 3-dimensional vector measurement in cartesian space indicating the measured variable magnitude and direction. These measurements create valuable information to investigate the robot motion characteristics and behavior. Also, it allows the roughly estimate the position of the robot inside the pipe with respect to time. The utilized approach is directly based on one of the most popular, modern data processing technique, deep learning. The main criteria on choosing an IMU is that the device utilized should be in low dimensional, low energy consumption. BNO055 is one the most popular, highly reachable, cheap IMU device and supports I2C, SPI and serial UART communication, which allows to configure the system to utilize in different conditions and several master data retrieval devices such as single board computers. Also, a serial connection instrument, FT232H via UART, is necessary to adapt the data flow port of IMU device. To construct a training dataset for a pipe line motion segment detection system, it is required to define a general pipe structure which the leak detection robot travelling inside. To make a general assumption for the proposed idea a pipe line system could be considered as constructed by finite geometric shapes. The structure simply constructed by straight segments and bending segments. From the travelling MICR's point of view, also from the core motion measurement device (IMU), the traveling path contains straight segments, right turning segments and left turning segments. Due to this finite state geometric constraint on a pipe structure, it possible to separate measurements taken from IMU device, which also represents the traveling path of the robot inside the pipe, into 3 different classes. This idea leads to generate a training data set containing these 3 classes, straight, left turn, right turn. The developed idea depends on the generation of artificial time graph images for the considered sensor measurement variables. The created time graphs (related to the measurement instance) successfully fed into the trained convolutional neural network model. The classifier is responsible to distinguish each individual measurement's motion type, straight, left turn, right turn, by analyzing the segment which takes the interested measurement point as its center. This approach allows the MICR to obtain knowledge, within time frame information, about the traveling action through a pipe line system. 2.Leakage signals Previously, hydrophones (i.e. microphone designed to be used underwater) were used to listen and record acoustic data (20- 20000 Hz) within water. This analog data was converted to digital using either commercial analog-to-digital converter sound card of computers. It was shown that the sound originated from leakages and cracks has low frequency (peak frequency is about 105 Hz) and there is almost no content above 5000 Hz. The upper frequency limit of sound signals coming from the cracks and leakage depends on the type of the pipe material, flow rate, and pressure. It was shown that the frequency range for Medium Density Polyethylene (MDPE) pipes is 20-250 Hz. It is known that leak signals rarely contain frequency components above 1000 Hz and 200 Hz in metallic and plastic pipelines, respectively (Hunaidi, 2006). It was measured that the amplitude of the signals diminishes very rapidly (0.25 dB/m) (Hunaidi, 1999). These outcomes were verified by utilizing the the planned microphone array (ReSpeaker Mic Array) and obtained satisfactory results which indicates signal strength and phase variation between different sound recording channels. However, as indicated about the important issue is to rebuild the microphone array within a tiny package. 3.MICR Prototype We developed a prototype containing a camera module, actuator and a propeller to observe the effectiveness of the actuator under water. This prototype additionally contains an inertial measurement unit to collect motion information to localize the robot. Other sensors and controller adaptation still in process. As indicated the devices should be resigned to make them smaller. Objective-2) Design and implement a control algorithm for MIRC Meso-Scale Intelligent Capsule Robots are becoming popular and they were not even in the consideration in robotics field 15 years ago. The inevitable issue on these kinds of tiny robotic system is to controller size and, of course, it's ability (control plan complexity or calculation power). Whenever a controller becomes smaller, it is proportionally converted into less capable version. Unfortunately, redesigning such a complicated microcontroller-oriented system consumes a lot of time. 1.Passive Control By utilizing passive motion control, the robot is capable of moving with the flow of the water inside the pipes. While the motion is performed, it measures the traveling information through the principle localization device (IMU) to investigate traveling sections through the pipe with respect the time. 2.Active Control By adapting the system to apply an active control, it is required to handle its motion inside a steady waterful region. A flagella-propelled bio-inspired submersible system is built and Its simulation results are successful. Similarly, a small dimensional controller must be adapted and utilized. Objective-3): Design and implement of an intelligent data management system We have almost finished data management system architecture and most of the component of the overall system is developed individually such as localization information analysis tool which works on inertial measurement unit measurements. Sound signal retrieval and offline analysis tool is also developed for microphone array device. The remaining sections are related to the real time application data which is going to be acquired after the robot hardware completed.
Publications
- Type:
Conference Papers and Presentations
Status:
Awaiting Publication
Year Published:
2019
Citation:
Application of Deep Learning to IMU sensor motion. Matthew Christian, Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Sambit Bhattacharya, Rashad Bailey, Kazuhiko Kawamura, S. Keith Hargrove. Application of Deep Learning to IMU sensor motion , Southeastcon, 2019, IEEE
- Type:
Conference Papers and Presentations
Status:
Awaiting Publication
Year Published:
2019
Citation:
Modeling And Testing Of Magnetic Speed Controlled Submersible Robot For Hydroponic Production. N.Turgay Tekin, Erkan Kaplanoglu, Erdem Erdemir, Fulya Baysal-Gurel, Cihan Uyanik and S.Keith Hargrovey. Modeling And Testing Of Magnetic Speed Controlled Submersible Robot For Hydroponic Production , Southeastcon, 2019, IEEE.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2019
Citation:
A deep learning approach for motion segment estimation for pipe leak detection robot. Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Ali Sekmen. A deep learning approach for motion segment estimation for pipe leak detection robot . 3rd World Conference on Technology, Innovation and Entrepreneurship, 2019, Elsevier
- Type:
Conference Papers and Presentations
Status:
Under Review
Year Published:
2019
Citation:
Deep Learning Approach for Motion Trajectory Estimation for Robotic Systems. Cihan Uyanik, Erdem Erdemir, Erkan Kaplanoglu, Sambit Bhattacharya, Rashad Bailey, Matthew Christian, Kazuhiko Kawamura, S. Keith Hargrove. Deep Learning Approach for Motion Trajectory Estimation for Robotic Systems . International Conference on Robot and Human Interactive Communication, 2019, IEEE
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2018
Citation:
Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection. Fulya Baysal-Gurel, Erdem Erdemir, Tim Darrah, Erkan Kaplanoglu, Robert Turner and S. Keith Hargrove, Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection , ICPP 2018
|
Progress 03/15/17 to 03/14/18
Outputs
Target Audience:Graduate assistantship: Since Fall-2017, one M.S. level and one Ph.D. level students participated into this project. One M.S. student successfully completed his proposal defense in April 2018. The students involved in protocol development, determining inoculation methods, experimental design, conducting experiments, data collections for the projects. Extension/Outreach:The research project has been mentioned in the university's annual research report and took attention of many students. I have now 2 EE, 1 ME and 1 CS students learning how to build these robots. We have also stated designing the course outline for the Objective-4. (Redesign of existing COMP/ENGR 4440 Mobile Robotics course). Also our lab is the most visited lab in the university and always our group has been given the task to do the lab tours to show the quality of research is being done in our university. (Some sample visits: Commanding General MG Cedric Wins from the U.S. Army Research, Development and Engineering Command, Division of Research and Institutional Advancement Nashville Chamber of Commerce Research and Innovation Tour and many high and middle school students) Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Two undergraduates are learning 3D Printing, one is learning PCB board design and soldering. We are working with these students and have already started redesigning mobile robotics course. How have the results been disseminated to communities of interest?We have shown the first results of our project in these two publications. Fulya Baysal-Gurel, Erdem Erdemir, Tim Darrah, Erkan Kaplanoglu, Robert Turner and S. Keith Hargrove, "Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection", ICPP 2018. (NIFA Support Acknowledged) Erdemir E. and Darrah T., "Real-Time Root Monitoring of Hydroponic Crop Plants: Proof of Concept for a New Image Analysis System", AETA 2017, Lecture Notes in Electrical Engineering, vol 465. Springer, 2018. (NIFA Support not Acknowledged) Also the first results of our research are mentioned in the annual research report of our university. What do you plan to do during the next reporting period to accomplish the goals?We will finish objective-1 next year. We have already started objective2,3 and 4. We will publish the first prototype designand results. Objective 1:Designing and implementation of a Meso-Scale Intelligent Capsule Robot (Year 1-2) We have finished %60 percent of this objective and will finish the rest next year. Objective-2) Design and implement a control algorithm for MIRC (Year 1, 2)We have finished %10 percent of this objective. we have now a working prototype and we will start working on the active control of the robot next year. Objective-3: Design and implement of an intelligent data management system (Year 2 & 3)We have finished %10 percent of this objective and will finish the rest next two years. We will start working on that part after we complete the first protype trials and start taking data from it. Objective-4: Redesign of existing COMP/ENGR 4440 Mobile Robotics course. (Year 3)We have finished %10 percent of this objective and will finish the rest next two years. Four undergraduate students started working in the lab voluntarily. We are teaching them the state-of-the-art technologies and tools. Meanwhile we have started the redesigning stage of the mobile robotics course.
Impacts
What was accomplished under these goals?
Objective 1) Designing and implementation of a Meso-Scale Intelligent Capsule Robot (Year 1-2) The aim of this study is to create a fast, accurate, and cost effective meso-scale intelligent capsule robot (MICR) to detect leakage in water pipelines. The proposed robot will be designed as an aerodynamically stable, light-weight, and waterproof robot to travel in water pipelines easily and steadier. It will consist of major sensors (e.g. microphones, digital camera, inertial measurement unit (IMU)) as well as microcontroller unit (MCU), lithium-ion rechargeable batteries, SD card, and 3D printed parts. The sensory information will be stored in SD card for post-processing (i.e., auto-interpretation, localization) by software algorithms. The robot's orientation and position data will be recorded with a mounted inertial measurement unit, which will measure angular rate in degrees per second, acceleration and changes in its magnetometers. The novel approach of this project is (1) to use MICR robots to inspect leakages in water pipelines and (2) to couple sensory data (acoustic, image) with IMU to localize the leakage. 1.1. Leakage signals Previously, hydrophones (i.e. microphone designed to be used underwater) were used to listen and record acoustic data (20-20000 Hz) within water. This analog data was converted to digital using either commercial analog-to-digital convertersor sound card of computers. It was shown that the sound originated from leakages and cracks has low frequency (peak frequency is about 105 Hz) and there is almost no content above 5000 Hz . The upper frequency limit of sound signals coming from the cracks and leakage depends on the type of the pipe material, flow rate, and pressure. It was shown that the frequency range for Medium Density Polyethylene (MDPE) pipes is 20-250 Hz. It is known that leak signals rarely contain frequency components above 1000 Hz and 200 Hz in metallic and plastic pipelines, respectively (Hunaidi, 2006). It was measured that the amplitude of the signals diminishes very rapidly (0.25 dB/m) (Hunaidi, 1999). It was also indicated that temperature and season is another factor that affects the leakage signals and their amplitude-distance relation. 1.2. Sound Localization Recent studies focusing on sound localization proposed various methods. Saxena et. al. designed an artificial pinna (outer ear) for sound localization (Saxena). Calculating interaural differences in time (ITD) and level (ILD) Guentchev et. al. showed that two microphones can be used to mimic sound localization in humans.They succeeded to reach 3 degrees angular error and 20% radial distance error. Nakano et. al. designed a "T"-shaped microphone array using four microphones to localize sound in three dimensions. The sound can be localized using three microphones if the sound source being detected is the only or dominant one. Thiemjarus et. al. proposed a Bayesian approach to localize the sound using three microphones placed on equilateral triangle. Previously proposed sound localization methods exploit either machine learning techniques (e.g. self-organizing map, active direction-pass filter, head-related transfer functions). 1.3. Experimental Setup To test the audio units and the robot, we are building a pipeline setup. The flow rate and pressure of the water within the setup will be adjustable from outside using compressors, pressure regulators, pressure and flow gauges, and valves. We will test different types and sizes of cracks within this setup by preparing different replaceable pipe segments. 1.4. Pilot Tests After the literature view, a basic hydrophone system was built using an electret microphone, canister, and a mineral oil to test selected electret microphone. The canister was filled with mineral oil and put the microphone within the canister. After sealing the canister, the hydrophone was ready to use. To receive and test the sound data, the hydrophone was directly connected to the microphone input of the computer's sound card. Using speech analysis program called Praat, developed by Paul Boersma and David Weenink, the sound data was recorded and analyze while the hydrophone was kept under water. Pilot tests showed that electret microphones are suitable to detect sound signal under water. The selected microphone is water proof and its frequency range is very high (20-20000Hz). The low frequency signals contribute much which is like the findings of the previous studies. We found that the major content is below 250 Hz. We tested preamplifiers of Texas Instruments, Analog Devices, and Maxim Integrated and successfully amplify audio signals as well as compress noise. We decided to use Maxim Integrated preamplifiers for audio unit. 1.5. First MICR Prototype We developed a prototype containing a camera module, actuator and a propeller to observe the effectiveness of the actuator under water. The white coverage is made from polymorphic material which melts and shapes easily in hot water. Analog audio unit receives audio signal from three microphones and process in analog units. Three signals are amplified using multichannel preamplifiers and noise are compressed. The amplified signals are transmitted to multichannel switch capacitor filter for anti-aliasing and amplified again using multichannel opamp to leverage the small signal to line level. Analog switch controlled by MCU transmits audio data package belonging to each microphone separately to MCU through analog input pin of MCU. Objective-2) Design and implement a control algorithm for MIRC (Year 1, 2) The development and application of active locomotion for Meso-Scale Intelligent Capsule Robot s have now reached a level of maturity where significantly enhanced control and inspection functionalities can be achieved for capsule robotics that were not available a decade ago and often thought impossible. This research is focusing on a passive control and active control of these robots inside the pipes. 2.1. Passive Control (Year 1) The first prototype of the robot control was a passive control in which the robot was moving with the flow of the water inside the pipes. The first results of this system are shown in the previous sections. 2.2. Active Control (Year 2) The objective of this year is to design and optimize a flagella-propelled bio-inspired submersible robot that can operate in hydroponic systems to assess the pipe zone health and operate in irrigation pipes to detect leaks by data fusion and image processing techniques. This system can be also used for the health assessment the roots in hydroponic farms. The research team designed a flagella-propelled submersible inspection robot and verified its maneuverability and operability under/around/through plant roots autonomously and through the irrigation pipes. Future studies will be on evaluation of robot control performance for determination of leaks in irrigation pipes.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2018
Citation:
Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection
Fulya Baysal-Gurel, Erdem Erdemir, Tim Darrah, Erkan Kaplanoglu, Robert Turner and S. Keith Hargrove, "Design of a flagella-propelled bio-inspired submersible robot for hydroponic production and irrigation system inspection", ICPP 2018.
|