Mapping Rabbit Burrows 2015

From Mindworks
Jump to: navigation, search
The Burrow Exploration Vehicle
Sponsor :
  • Dr. Janet Rachlow
  • Dr. Courtney Conway
Team Name : Vespucci
Duration : Fall 2015 - Spring 2016
Faculty Advisers
  • Dr. Joel Perry
  • Bruce Bolden
Students
  • Xihua Chen (EE)
  • Stephen Hanes (EE)
  • Tyler Jaszkowiak (CS)
  • Brett Menzies (CS)
  • Lance Wells (CS)
  • The pygmy rabbit (Brachylagus idahoensis) is a North American rabbit, and is one of only two rabbit species in America to dig its own burrow. An isolated subspecies in the Columbia River Basin is listed as endangered by the United States.[1]. The burrowing owls of North America, who raise their young in burrows dug by other animals[2], are also of interest to researchers at the University of Idaho. Current methods for exploring burrows are either ineffective or invasive, with the leading tool for exploration being a camera attached to a single flexible rod. This limits the type of structures the device can enter and the depth it can reach. The goal of this project is to design and build a device that can enter a burrow to collect video, volume measurements, and a 3D map of the structure. The current iteration of the project aims to write a software interface for controlling the vehicle designed in 2014-15 and collecting data.



    Problem Definition[edit]

    The goal of this project is to design a system for controlling and collecting video & spatial data from the existing burrow-exploring vehicle.

    Background[edit]

    Pygmy rabbits dig burrows into the ground, which help to keep them safe from predators and the cold. These burrows also affect the soil density, temperature, and water retention of the soil. Analyzing the structure of these burrows can lead to a deeper understanding of these issues as well as the lives of pygmy rabbits. Burrowing owls raise their young in structures dug by other animals that are equally important to understand.

    Burrows can be difficult to access by conventional means. The best current method involves using a camera attached to a flexible rod. When more information is needed, digging can be utilized, but is destructive to the animals' habitat. The following properties are believed to be typical of pygmy rabbit burrows, and demonstrate the accessibility challenge.

    • Burrows have a diameter between three and eighteen inches.
    • On average, burrows have typically between three and five entrances, with an expected maximum of fourteen entrances.
    • Burrows are between a meter and one and one-half meters under ground.
    • Each burrow entrance will be between two and three meters away from other entrances.

    2014 Vehicle Specifications[edit]

    Parameter Metric Acceptable Value Target Value Actual Value
    Device Size Diameter in inches 3" 3.5" 3"
    Minimum Sensor Distance Millimeters 10mm 0mm 0mm
    Maximum Sensor Distance Centimeters 10cm 20cm 10cm
    Top Speed Meters per Second 0.25 0.25 0.25
    Video Quality Resolution 640x480 1280x720 1280x720
    Run-Time hours 1 1 1
    On-Board Data Memory Giga Bytes(GB) 4 8 16
    On-Board Video Memory Giga Bytes(GB) 32 32 32

    The requirements for the project are as follows.

    The vehicle must:

    • Fit inside wide range of rabbit burrows.
    • Take measurements of the burrow space which may be used to provide volumetric data.
    • Record images of the burrow interior to be used by client.
    • Record all relevant data and return to the surface in a timely manner.
    • Be usable by the client with no special skills.
    • Operate for a period of one hour.

    2015 Device Specifications[edit]

    Priority Parameter Metric Acceptable Value Target Value Actual Value
    Very High Robot Control Robot control accessibility The robot can be controlled via user interface. The robot can be controlled via user interface. Complete
    Very High Video Quality Video Quality The video quality is sufficient to navigate burrows and showcase to interested parties. The video quality is sufficient to identify individual eggs and owl characteristics (including age). Sufficient to read text at a distance.
    Very High Camera Functionality Camera video feed accessibility The interface can display video feed from both cameras interchangeably. The interface can display video feed from both cameras interchangeably. Single Camera
    High Camera Visibility Range 6 in. 12 in. 36 in.
    High Endurance Operating time on a single battery charge 1 hour of intermittent use 1 hour of continuous use 3 hours of heavy use
    High Control Ease-of-Use Control accessibility The controls are functional and require little effort to learn. The controls are functional, easy to learn, and self-explanatory. 4 control keys
    High Maneuverability Average Response Time <100ms <10ms ~100ms
    Low Robot Charging Charging accessibility Charging can be accomplished without opening the vehicle case. Charging can be accomplished in the field. Chargeable while assembled
    Low Camera label interface Camera label accessibility The interface accurately labels the active camera. The interface accurately labels the active camera. Single camera
    Very Low Mapping Functionality Generated map quality The included software is capable of generating a line map of the vehicle's path. The included software is capable of generating a 3D map of the portions of the burrow explored. No mapping
    Very Low Mapping Interface Mapping function accessibility The software can generate a map by post-processing data. The software can generate a map in real-time while driving. No mapping
    Very Low Volume Estimate Relative error No volume estimate. <10% relative error. No mapping
    Very Low Control Device Program accessibility The software can operate on a laptop. The software can operate on a tablet or a laptop. Any computer with 2 USB ports

    Project Learning[edit]

    2014 Project State[edit]

    Objective: To determine the state of the project as it was handed to us in September 2015, after the completion of the previous iteration in May 2015.

    Methodology: Each team member carefully examined last year's wiki page and the contents of the dropbox folder we received from the clients containing work from last year's group.

    Findings and Analysis:

    • The designed method for transmitting video from the vehicle is an analog signal over a coaxial video cable.
    • A composite to USB converter is intended for use in reading this signal on a computer in real-time.
    • The full quality video from the forward camera is also recorded to the on-board DVR.
    • It is unclear how the rear camera footage is transmitted.
    • The specifications for the distance sensors chosen indicate that their maximum range is 10cm. This could pose a problem for burrows up to 18in. (46cm) wide or at junctions.
    • The computer in the vehicle has a 454 MHz ARM processor and 64 MB of RAM.
    • The following hardware items had not yet been addressed and were included on the "To-Do" List given us:
      • Mount LED light source
      • Construct a durable, keyed tether for burrow exploration
      • Attach accelerometer
      • Mount range finder array
      • Purchase a connector for battery charging
      • Wire in begin and stop recording commands, utilizing the existing button on the DVR
      • Identify the correct camera failure modes
      • Install an external power switch.
    • In addition, we noted that the forward camera was not wired to anything.
    • The following software items were included on the "To-Do" List:
      • Create a comprehensive user interface
      • Write code for sensor data storage
      • Implement a method to create a 3D map from points obtained by sensor readings
      • Implement a method for estimating burrow volume
    • The file "Tech Presentation Final" suggests black and white 640x480 video is captured, whereas the wiki indicated 1280x720 video.
    • Battery recharges in about 1.5 hours from "almost any power source", though it's not clear what that means, since the "To-Do" List included purchasing an adapter for charging.
    • The existing C code implemented keyboard controls for the motors that were not simple or easy to learn/remember.

    Range Sensor Analysis[edit]

    Objective: Determine the accuracy, range, and electrical characteristics that can be expected from the STMicroelectronics VL6180X Proximity sensors purchased by the 2014-15 team.

    Methodology: Analyse datasheet and perform manual tests

    Findings and Analysis:

    Range Sensor Performance[3]

    The observed maximum sensor range was about 15cm. In cases where a burrow is 18in. (46cm) wide, this will still not reach halfway across the tunnel. However, this is significantly better than the 10cm. range suggested by the specifications from the vendor. The small maximum range may still negatively impact the vehicle's ability to create maps and estimate burrow volume. The minimum accurate range appears to be 1cm. This should be acceptable most of the time, since only some portions of the burrows should be so small that the vehicle can barely traverse through them. Additionally, the error at distances that small should not significantly impact the relative error of volume estimates. Lower light levels seemed to improve performance. Strong, directed light sources at a low angle of incidence could drop the maximum range to 5-6 cm.

    A range sensor is capable of transmitting a reading 100 times per second.

    The distance sensors all communicate using the same predefined address over the I2C serial protocol; the address conflict will present challenges when we try and wire more than one together. They also operate at a lower voltage than the normal logic levels used in the robot hardware. Our electrical team will need to prepare some custom circuitry to drive an array of these sensors.

    Design[edit]

    Electrical[edit]

    Battery[edit]

    Replacement Battery
    Front Camera with converter and Memory

    In the original design, the singular port at the rear of the vehicle was used both for communication with the vehicle during operation and for charging of the battery. This limited the number of connections that could be used to transmit data and video, since a number of wires on the port were required for the charging circuit. The redesign has an additional port in the rear of the vehicle's lid for charging. This saves connections on the main port for transmission of data and video.

    The original design used an 11.1V Lithium-ion 3-cell 1350 mAh Prolite Power battery. During storage, it had decreased below minimum voltage. The replacement battery purchased is similar in performance and slightly smaller in size.

    Camera[edit]

    The vehicle, as designed in 2014-15, has fore and rear cameras. The front-facing camera saves video to an on-board memory card. However, for capturing and transmission to be activated, a small button must be toggled on the chip. After examining the circuit, we hope that the switch can be triggered electrically. The same must be done for the switch that toggles between the fore and rear camera for transmission.

    Position Tracking[edit]

    Optical Flow Sensor
    In order to create a map of a burrow, the vehicle must have a mechanism for determining its own position relative to where it began. The distance traveled by the vehicle is not equivalent to the amount the motors have turned, however, since any tread system requires the treads to slip in order to turn. This will introduce errors in the position estimate. We intend to use an optical flow sensor, as found in computer mouses, to track distance traveled. Combined with data from the gyroscope and accelerator, this information should be adequate to estimate a relative position.

    The chassis of the vehicle already sits low enough to the ground to make an optical flow sensor effective. However, it must be determined if the coarseness of the soil in burrows or the lighting will provide difficulties and additional sources of error. If so, using data from the motor encoders combined with data from the optical flow sensor may be desired. Or, if this would not be more accurate than simply utilizing the motor encoders, the optical flow sensor may not be necessary.

    The optical flow sensor was installed, but the current version of the software does not utilize it.

    Range Sensors[edit]

    As in the original (though unimplemented) design, we plan to mount five range sensors around the hull of the vehicle. These will measure the distance from the vehicle body to the wall of the burrows at various angles from 0 to 180 degrees. Below the vehicle, the area of the burrow is assumed to be essentially constant.

    Unknown to the previous team, the purchased range sensors all share a single address. In order to distinguish them, we will require an I2C multiplexer. This additional piece of hardware will allow us to treat each sensor as a separate entity for control and data processing in the software.

    The range finders are not implemented in the final version of the robot.

    Schematic[edit]

    File:2015-VisionMapping-Schematic.pdf The final wiring layout of the robot, as-built. The battery (not shown) is the source of both ground and +11.4V.

    New Functions[edit]

    Per client request, the final version of the robot includes some extra features.

    • The headlight can be manually toggled from the driver station.
    • The motor speed can be controlled by a slider from the driver station.
    • The arduino monitors the battery voltage while the robot is on and transmits status to the driver station. The driver station displays the status as a traditional battery level icon.

    Software[edit]

    Communication[edit]

    Diagram showing how components communicate to achieve objectives.

    The vehicle transmits video via a composite video cable. This signal goes through the StarTech device for conversion to USB, which is then plugged into the client computer. The client computer will recognize the vehicle as a webcam, so the control software can display it in real-time.

    Data about the location of the vehicle and points on the burrow wall around it are transmitted via a separate line. This line also carries control commands from the client software to the vehicle. Messages are comprised of a byte-long header, followed by a byte indicating the type of message. Several bytes of payload data follow. Finally, a checksum is appended, which is just the XOR of the previous bytes. A byte-long footer indicates the end of a message. The checksum allows for messages that have been corrupted in transmission to be rejected.

    Messages include:

    • MotorCommand
    • AccelData
    • PositionData
    • Error
    • Headlight

    User Interface[edit]

    The user interface design is intended to make control of the vehicle as simple as possible. One primary objective of design is to give the video feed from the vehicle as much screen space as possible. However, we recognize that the video transmitted is 720p HD and that many laptop screens will be larger than this; some screens may allow for additional real estate to be used for other purposes.

    The interface consists of three tabs. The first shows primarily live video from the vehicle, and also controls for recording, switching cameras, and controlling the headlights. The second tab shows a map of the explored burrow and an estimate of its volume. The third tab shows messages passed between the computer and the vehicle, for debugging purposes.

    Personnel Information[edit]

    Team Members[edit]

    Picture
    2015 RabMap Team.jpg
    Team Members (Listed from Left to Right) Discipline
    Stephen Hanes: Stephen is working towards a Bachelor of Science in Electrical Engineering at the University of Idaho. His main focuses are antennas and AC power, but he also has experience with microcontrollers and digital logic. He chose this project in hopes of applying his field experience. EE
    Xihua (Jake) Chen: Jake, a senior, is pursuing a Bachelor of Science in Electrical Engineering at the University of Idaho. He interest of study are microelectronics and semiconductor devices. He selected this project to apply his knowledge of electronics in order to make the robot functional. EE
    Tyler Jaszkowiak: Tyler is a pursuing Bachelor of Science degrees in Mathematics and Computer Science at the University of Idaho. His interests in the field are primarily artificial intelligence and statistical machine learning. He chose this project in order to solve interesting problems in Human-Computer Interface design and robotics. He is involved on campus both as a member of the Homecoming Committee and an executive in his fraternity. CS
    Brett Menzies: Brett is attending University of Idaho to earn Bachelor of Science degrees in Computer Science and Mathematics. His interests are in Robotics, Software Development, and Algorithms. He chose this project to make use of his relevant experience and out of an interest in the unique challenges provided by spatial data acquisition and mapping. CS
    Lance Wells: Lance is a Computer Science Major at the University of Idaho in his fourth year. His research interests include user interface design and graphical implementation in software. He chose this project in attention to interface design and communicative space mapping. CS

    .

    Clients[edit]

    Dr. Courtney Conway received a B.S. in Wildlife Biology from Colorado State University, an M.S. in Zoology from the University of Wyoming, and a Ph.D. in Organismal Biology and Ecology from the University of Montana. He works on applied questions to aid wildlife managers make informed decisions, and also works on basic questions to better understand the ecological processes that affect behavior and demography of animal populations.

    Dr. Janet Rachlow is a mammalian ecologist interested in behavior and conservation of both rare and common mammals. Her current research focuses on habitat relationships of diverse species, with an emphasis on understanding the consequences of habitat modification. Rachlow along with her students and collaborators conduct field and laboratory studies to address questions that can help manage and conserve wildlife and their habitats. Janet enjoys outdoor activities and Idaho’s terrific wildland resources.

    Instructors[edit]

    Bruce Bolden received his bachelor's degree and master's degree in mechanical engineering from the University of California, Davis. His graduate research involved simulation of the Electron Beam Welding process. Following graduation he worked in the aerospace industry for several companies before joining a small mathematical software company in 1990 as the Engineering Manager. Prior to joining the University of Idaho in 1997, Bruce worked on the development of several Microsoft Windows applications.

    Dr. Joel C. Perry is a new member of the Mechanical Engineering Department as an assistant professor with a focus on robotics and engineering design. He received a B.Sc. degree in mechanical engineering from Gonzaga University in 2000, and M.Sc. and Ph.D. degrees in mechanical engineering from the University of Washington, in 2002 and 2006 respectively. Dr. Perry spent the past 6 years working abroad in the Department of Rehabilitation Technologies at Tecnalia Research & Innovation in San Sebastian, Spain, where he managed R&D activities in the development of low-cost solutions for upper extremity rehabilitation. Before joining Tecnalia, Dr. Perry was involved in the development of a 7 degree-of-freedom (dof) arm exoskeleton, a 5-dof high precision positioning robot, a 5-dof surgical simulator, a novel 2-dof surgical grasper, and a 1-dof powered prosthesis for early-stance gait improvements in trans-tibial amputees. His research interests include enabling technologies for upper and lower limb disability, rehabilitation robotics, and surgical robotics.

    Document Archive[edit]

    References[edit]

    1. Pygmy rabbit - Wikipedia[1]
    2. Burrowing Owl - Wikipedia[2]
    3. VL6180x datasheet - STMicroelectronics[3]