McGill Autonomous Robotics Team (M.A.R.T.)
McGill Autonomous Robotics Team (M.A.R.T.)
MARTy 2.0: Moving Forward s Abstract MARTy 2.0 is an Autonomous Underwater Vehicle designed and constructed by a group of students at McGill University. It is a completely new design, after several aspects of the previous revision proved to be problematic. The center of the new design is a rectangular hull constructed from sheets of aluminum. This provides better water sealing, easier access to equipment, and ideally placed windows. Additionally, the hardware and software which make up the electronic control system, have been revised and expanded. Authors Richard Cohen, Philip Davidson, David Goldbaum, Matthew Guttman (Team Lead), Christopher Harvey, Ghalia Lahlou, Matthew Michini, Svilen Savov, Alan Schoen, Vincent Thompson (Faculty Supervisor)
Introduction MART’s goal is to complete several objectives in the underwater robotics competition held by the Association for Unmanned Vehicle Systems International (AUVSI) and the Office for Naval Research (ONR). In 2010, we will attempt to pass through the validation gate, follow the path, and complete the life vest and hedge tasks. Foundations for the New Design
CAD image of final design
In 2009, MARTy 1.0’s hull was not watertight. This resulted in the loss of some electronics and prevented the use of other vital electronics for fear of damaging them as well. This experience shaped our design decisions for the second revision: MARTy 2.0. Hull The central hull is composed of a 17”X19”X11” rectangular box. Most of this box is composed of 1/8” aluminum sheeting, which was cut, bent and welded to the desired dimensions. Two Lexan windows, on the front and bottom of the hull, are used as camera ports. The front window is large to allow room for the future addition of cameras and to allow certain electronics to be viewed
M.A.R.T.
from outside the hull.
Front view of AUV showing the large window The back of the hull is open, and can be sealed with a custom aluminum endcap. The endcap is machined from a solid 19”X11”X7/8” piece of aluminum. It has 10 pass-‐through ports for connecting external electronic components. The hull is sealed by compressing an o-‐ring between the endcap and the hull. 12 screws distributed around the perimeter ensure even and constant distribution of pressure.
Detailed view of endcap
External components are mounted on a frame built from 80/20 aluminum bars. This allows modular components to be added, removed
2010 Journal Paper
2
and adjusted dynamically. Several electronic parts are mounted on this frame: thrusters, battery modules and the depth sensor. Batteries are housed in sealed compartments built from ABS tubes. Each holds 6 NiMh D-‐Cell batteries. A single 2-‐conductor wire connects each to the main hull
The electronics rack, with endcap and o-ring visible
External battery compartment
MARTy uses to kinds of thrusters. Two Seabotix BTD150 thrusters provide forward and backward thrust. Six additional custom built thrusters enable motion in all other directions. The custom thrusters are built from modified Rule bilge pumps. The internal electronics are mounted on a rack, which is affixed to the removable endcap. The rack is equipped with plastic sliding rods to prevent friction with the hull so it can be removed easily. Boards are mounted on a large fiberglass sheet which is affixed to the rack.
M.A.R.T.
Sensors MARTy features two logitec quickcam cameras positioned to allow frontward and downard vision. One Keller Levelgage depth sensor measures the altitude. The 4-‐20 mA output is read by an AVR microcontroller. A Memsense nIMU measures the heading of the AUV. Power System Power is provided by 12 NiMH D-‐Cell batteries. The battery array provides 17-‐20V power. A single PicoPSU provides the voltages required by electronics. Power to the thrusters is adjusted dynamically with Dimension Engineering Sabertooth motor controllers. Computation and Communication The main software system is housed on a Lippert Cool Roadrunner 4 embedded computer. The computer runs an operating system based on Debian Linux. USB is used for communication between the main board and an array of microcontrollers, which manage the AUV’s assets, such as motor
2010 Journal Paper
3
controllers, and the depth sensor. The one exception is the IMU, which communicates through a serial connection. The main software is implemented in C++ with extensive use of the Boost libraries. In addition the Eigen math library is used for matrix computations. OpenCV is used for the visual system. We have implemented visual strategies for the path, life vest, and hedge tasks. Each uses a similar strategy. Objects are segmented by grouping pixels according to their color values in CYMK space. Ellipse fitting is used to determine the size and orientation of objects. A variety of noise-‐reduction and information-‐ isolation strategies are used in different stages of the tasks.
Frame from the line-following task processed with our visual system
during the construction of our hull. We also thank all of our sponsors. Gold: McGill Engineering Silver: Lippert, Future Electronics, McGill Engineering Undergraduate Society Bronze: Memsense, Advanced Circuits, Digi-‐Key, TSlots, Anica Steel, Fastenal, Johnston industrial plastics, Verdun Anodizing, Keller America
Acknowledgements We would like to thank the McGill Faculty of Engineering for the generous support, the Department of Mechanical Engineering for their administrative support and the McGill Engineering Design Network for providing space. We thank the staff members of the machine shops at McGill for their time and knowledge, which was incalculably valuable
M.A.R.T.
2010 Journal Paper
4
MARTy 2.0: Moving Forward s Abstract MARTy 2.0 is an Autonomous Underwater Vehicle designed and constructed by a group of students at McGill University. It is a completely new design, after several aspects of the previous revision proved to be problematic. The center of the new design is a rectangular hull constructed from sheets of aluminum. This provides better water sealing, easier access to equipment, and ideally placed windows. Additionally, the hardware and software which make up the electronic control system, have been revised and expanded. Authors Richard Cohen, Philip Davidson, David Goldbaum, Matthew Guttman (Team Lead), Christopher Harvey, Ghalia Lahlou, Matthew Michini, Svilen Savov, Alan Schoen, Vincent Thompson (Faculty Supervisor)
Introduction MART’s goal is to complete several objectives in the underwater robotics competition held by the Association for Unmanned Vehicle Systems International (AUVSI) and the Office for Naval Research (ONR). In 2010, we will attempt to pass through the validation gate, follow the path, and complete the life vest and hedge tasks. Foundations for the New Design
CAD image of final design
In 2009, MARTy 1.0’s hull was not watertight. This resulted in the loss of some electronics and prevented the use of other vital electronics for fear of damaging them as well. This experience shaped our design decisions for the second revision: MARTy 2.0. Hull The central hull is composed of a 17”X19”X11” rectangular box. Most of this box is composed of 1/8” aluminum sheeting, which was cut, bent and welded to the desired dimensions. Two Lexan windows, on the front and bottom of the hull, are used as camera ports. The front window is large to allow room for the future addition of cameras and to allow certain electronics to be viewed
M.A.R.T.
from outside the hull.
Front view of AUV showing the large window The back of the hull is open, and can be sealed with a custom aluminum endcap. The endcap is machined from a solid 19”X11”X7/8” piece of aluminum. It has 10 pass-‐through ports for connecting external electronic components. The hull is sealed by compressing an o-‐ring between the endcap and the hull. 12 screws distributed around the perimeter ensure even and constant distribution of pressure.
Detailed view of endcap
External components are mounted on a frame built from 80/20 aluminum bars. This allows modular components to be added, removed
2010 Journal Paper
2
and adjusted dynamically. Several electronic parts are mounted on this frame: thrusters, battery modules and the depth sensor. Batteries are housed in sealed compartments built from ABS tubes. Each holds 6 NiMh D-‐Cell batteries. A single 2-‐conductor wire connects each to the main hull
The electronics rack, with endcap and o-ring visible
External battery compartment
MARTy uses to kinds of thrusters. Two Seabotix BTD150 thrusters provide forward and backward thrust. Six additional custom built thrusters enable motion in all other directions. The custom thrusters are built from modified Rule bilge pumps. The internal electronics are mounted on a rack, which is affixed to the removable endcap. The rack is equipped with plastic sliding rods to prevent friction with the hull so it can be removed easily. Boards are mounted on a large fiberglass sheet which is affixed to the rack.
M.A.R.T.
Sensors MARTy features two logitec quickcam cameras positioned to allow frontward and downard vision. One Keller Levelgage depth sensor measures the altitude. The 4-‐20 mA output is read by an AVR microcontroller. A Memsense nIMU measures the heading of the AUV. Power System Power is provided by 12 NiMH D-‐Cell batteries. The battery array provides 17-‐20V power. A single PicoPSU provides the voltages required by electronics. Power to the thrusters is adjusted dynamically with Dimension Engineering Sabertooth motor controllers. Computation and Communication The main software system is housed on a Lippert Cool Roadrunner 4 embedded computer. The computer runs an operating system based on Debian Linux. USB is used for communication between the main board and an array of microcontrollers, which manage the AUV’s assets, such as motor
2010 Journal Paper
3
controllers, and the depth sensor. The one exception is the IMU, which communicates through a serial connection. The main software is implemented in C++ with extensive use of the Boost libraries. In addition the Eigen math library is used for matrix computations. OpenCV is used for the visual system. We have implemented visual strategies for the path, life vest, and hedge tasks. Each uses a similar strategy. Objects are segmented by grouping pixels according to their color values in CYMK space. Ellipse fitting is used to determine the size and orientation of objects. A variety of noise-‐reduction and information-‐ isolation strategies are used in different stages of the tasks.
Frame from the line-following task processed with our visual system
during the construction of our hull. We also thank all of our sponsors. Gold: McGill Engineering Silver: Lippert, Future Electronics, McGill Engineering Undergraduate Society Bronze: Memsense, Advanced Circuits, Digi-‐Key, TSlots, Anica Steel, Fastenal, Johnston industrial plastics, Verdun Anodizing, Keller America
Acknowledgements We would like to thank the McGill Faculty of Engineering for the generous support, the Department of Mechanical Engineering for their administrative support and the McGill Engineering Design Network for providing space. We thank the staff members of the machine shops at McGill for their time and knowledge, which was incalculably valuable
M.A.R.T.
2010 Journal Paper
4
