2016 VEX WORLDS ENGINEERING NOTEBOOK

Transcription

1 2016 VEX WORLDS ENGINEERING NOTEBOOK ODU VEX U Robotics APRIL 20, 2016 DR. GENE HOU & DR. CHUNG-HAO CHEN Allan Cahill, Timothy Clarke, Michael Darnell, Andrew Dearhart, Alfredo Delos-Santos, Darryl Sampson, Matthew Staley, Richard Stinson, Daquan Styles

2 Table of Contents List of Figures... ii List of Tables... ii Abstract... iii Introduction... 1 Background and Literature Review... 1 Methods... 2 Purdue Qualifier Robots... 3 Large Robot (First Iteration)... 3 Small Robot (First Iteration) CSM Skills Challenge Robots Skills Challenge Robot (First Iteration) Large Robot Changes (Second Iteration) World Championships Robots Large Robot (Third Iteration) Small Robot (Second Iteration) Skills Challenge Robot Changes (Second Iteration) Robot Programming Autonomous Driver Control Volunteering High School VRC Events Larchmont Elementary School STEM Night and 2016 Old Dominion University Admitted Students Day Freshman ENGN Homecoming Parade IEEE Southwest Conference Virginia Beach Aquarium Results Discussion Conclusion Appendices Appendix A Robot and MATLAB Code i

3 Appendix B Gantt Chart References List of Figures Figure 1 Initial x-drive Chassis... 4 Figure 2 Final Large Robot Construction... 6 Figure 3 Initial Launching Mechanism... 7 Figure 4 First CAD Design of Flywheel... 9 Figure 5 Lift with Small Robot On Top Figure 6 Lift Mounted to Back of Robot Figure 7 Flywheel and Hood of Small Robot Figure 8 Spiral Cam Equations Figure 9 Four Revisions of Cam Figure 10 Cam Launcher Mounted to Support Structure Figure 11 Intake Changes made to Large Robot Figure 12 Flywheel Changes for Worlds Large Robot Figure 13 Lift Mounted on Worlds Large Robot Figure 14 Drive System for Worlds Small Robot Figure 15 Passive Intake on Front of Chassis Figure 16 Hood on Worlds Small Robot Figure 17 Flywheel Mechanism for Worlds Small Robot Figure 18 Old (left) Ball Holder Compared to New (right) Figure 19 Andrew and Allan Inspecting Robots at the Hampton Roads VRC Event Figure 20 Richard and Mike Serving as the Head Referee and Emcee Figure 21 Richard Volunteering at the Larchmont Elementary School STEM Night Figure 22 Homecoming Robot on top of the join Engineering Float Figure 23 Gantt Chart List of Tables Table 1 Large Robot Drive Comparisons... 4 Table 2 First Launcher Range and Distance Tests... 8 Table 3 Small Robot Drive Comparisons Table 4 Small Robot Launching Rate Comparisons Table 5 Large Robot Launching Rate Comparisons Table 6 VEX_AllCalculations.m, Calculations Done for the 2015/2016 VEX U Team Table 7 LargeRobot_Comp.c, Large Robots Competition Code Table 8 SmallRobot_Comp.c, Small Robots Competition Code Table 9 SkillsRobot_Skills.c, Skills Robot Code ii

4 Abstract The VEX robotics organization works to educate and further the development of robotics at the college level. The objective of this project is to experimentally develop a minimum of two robots for use in the 2015/2016 VEX U Nothing but Net tournament. These robots were created through holding design meetings, prototyping, coding, and testing. Three final robots were built for this competition. Through a semester of testing and coding, an omnidirectional drive system was found to be the easiest drive system to code around for autonomous functions. A variety of flywheels were 3D printed for the launching mechanisms on the competition robots; the large robot launches at a velocity of mph, while the small robot launches at mph. A third robot was developed specifically for the programming skills challenge, which features a spiral cam launcher that has perfect accuracy and precision when placed on the field. The robots were built to perform both autonomously and operator controlled on a competition field, with the intent to place number one in the world, and thus receive an invitation to the world championships, for its programming scores. iii

5 Introduction The VEX U Robotics Team serves as home to a competition based robotics organization that can compete at and against colleges across the world, with some of the best engineering programs. During the previous year s competition, Skyrise, the team competed for its first year, placing 4th in competition at the College of Southern Maryland, beating out an old school rival, Virginia Tech [1]. This project is crucial to Old Dominion University because it provides an opportunity to earn prestigious awards and respect in a relatively new robotics competition. It allows for students, from Freshman to Seniors, to find a project where they might feel at home, surrounded by a tight knit competitive team looking to rise above other universities and engineering programs. The project is also very important to the club it was built around, as the success of the project gives the club a broader base to work with, as well as a potential for more financial support. Background and Literature Review The 2015/2016 VEX U Robotics Competition, Nothing but Net, takes place on a 12 ft square game field, played by two teams composed of students from universities around the world. Each team builds, at minimum, two robots for the competition; one of these robots must fit inside of a 24 in cube, and the other inside of a 15 in cube. The matches start out with a 45-second autonomous period [2] where the robots pretty much control themselves, void of any feedback from operators. After this autonomous period, a winner is determined for a match bonus, and a 75-second operator control period [2] begins, where two human drivers take control of the robots to score as many points as possible. To score in a match, the robots must score the game objects, 4-inch diameter foam balls, into either the low goal, a trapezoidal area of foam field tiles in the opposite corner from the robot s starting position, or the high goal, a 36-inch fiberglass and mesh structure above the low goal [2]. During the last 30-seconds of the operator control period, robots may lift their partner to score an additional bonus, potentially securing a win. 1

6 Every April, the VEX organization hosts the world championships for high school and college teams around the globe. In order to obtain an invite to the world championships, each team must either: win a regional tournament with ten or more teams, win the excellence award, the most prestigious award given out at the VEX U competitions, or place in the top three in a global leaderboards of the skills challenge rankings. The skills challenge, unlike the standard tournament style matches, allows a single team to place their robots on the field alone and tasks them with scoring as many points as they can in 60-seconds [3]. The Skills challenge is divided into two portions: the robot skills challenge and programming skills challenge. The robot skills challenge has drivers take control of the robots for the 60-second time frame. The programming skills challenge is entirely autonomous and doesn t allow human input for the 60- second time frame. A key design constraint in the VEX U competition is that only official VEX parts are permitted for use on the robot, with only minor deviations: each robot is permitted to use a single 12 by 24 in sheet of Lexan plastic, an unlimited number of 3D printed parts can be used as long as each part is smaller than 6 by 6 by 3 in, and each robot is only allowed to use a maximum of twelve motors [4], so that teams with a larger budget can t get a slight advantage with using more motors. Methods Throughout the competition season, the team went to a regional qualifier and a skills challenge event. Initial design concepts were used for the Purdue Winter Vortex VEX U Qualifier, while a completely new small robot was designed and built for the CSM Skills Challenge. Initially, work was split into the three main components of each robot for the game, Nothing but Net: a drive system, an intake device, and a launching mechanism. From these main components, designs were tossed around between the team members, while they were prototyped, tested, and eventually compiled together for the competition robots. While the 2

7 robots were being built, a smart field-centric programming model was uploaded to the Cortex Microcontroller. Purdue Qualifier Robots The Purdue Qualifier was a competition scheduled for February 20, 2016, and consisted of 10 teams from across the country. ODU was able to spend roughly a semester budgeting, designing and building robots that would allow the team to compete a high level. Large Robot (First Iteration) The initial plan for this robot was to shoot the majority of the driver controlled loads from the back corner, a distance of around 16 ft, and then play the rest of the field with the small robot. Drive System The initial design for the large robot drive system was an x-drive, shown in Figure 1, which would have allowed for movement in any direction, forwards and backwards, sideways, diagonally, and rotationally. This drive system is a fast, maneuverable, and ideal system for an operator to control, as seen in Table 1. Although it is very fast and maneuverable, the omnidirectional wheels used for the drive have a lot of built in slippage due to the small rollers along the edge of the wheels; these inconsistencies caused most autonomous tasks to get off track. 3

8 Figure 1 Initial x-drive Chassis Due to the autonomous tasks going off track, the ECE team wanted to instead use a mecanum wheel drive, which allows for the same type of movement an x-drive does, but has less slippage in the wheels due to a slightly higher coefficient of friction. This newly designed chassis and drive system was much slower, as shown in Table 1, but allows for a higher consistency in the autonomous programming, giving the robot a potential to outscore its opponents during the first 45-seconds of the match, securing the autonomous bonus. From an operator controlled perspective, the mecanum drive is not as effective, but a compromise had to be made for the sake of autonomous control. Table 1 Large Robot Drive Comparisons Drive System Time A (s) Time B (s) Average Time (s) x-drive Total Average (s) Standard Deviation Holonomic / Mecanum Total Average (s) Standard Deviation Percent Difference (%)

9 Intake Device An initial prototype design for the large robot intake was built around the same time as the initial prototype drive system and launcher; this prototype used flaps attached to a tank tread system to pick up the game objects. The intake was extremely slow and would stall if multiple game objects were in it. Due to this, a more compact system had to be designed and built around the drive system. After a drive system had been chosen for the large robot, construction began on the new iteration of the intake. Rather than using vertical rotation, the new design ended up using horizontal rotation from the drive motors. Two motors, with an internal gearing of 240 RPM, drove the same shaft, with a chainand-sprocket system driving additional shafts, geared for a ratio of 1:2. This allowed for a large section of the 24-inch cube to be used for the intake. A number of flaps were added to the chains, giving the intake a way to grip the game objects, making sure they would not leave the system. Similar to the initial design, the intake was slow, but it no longer locked up when multiple game objects were in it. At the back of the system was a large metal plate, where one side of the game object would be in contact with it while the other side was in contact with the flaps on the chain-and-sprocket system. This allowed for less motors to be used, keeping the number of total motors low, but also lowered the consistency of the intake. Dead zones, a section of the intake where the game objects get stuck, appeared in multiple places, forcing the design to potentially be reconsidered; however, rather than completely redesigning the intake, a solution was found by placing a small piece of aluminum to add a ramp in the section of the intake where the objects were getting stuck. Even with a fairly consistent intake, an issue arose when the ECE team attempted to code the autonomous tasks: the current design didn t allow for very many game objects to be picked up at once, especially not the small pyramid shaped stacks of objects placed around the field. A new front section of the intake that spun at a very fast rate was placed about 6-inches above the field tiles, driven by a chain from the main section of the intake. The main section designed previously was moved back about 5-inches from the front, closer towards the center. These two changes allowed for two or three of the pyramid 5

10 shaped stacks to be picked up consistently, more than the single game object that it could pick up previously. The largest inconsistency in the intake was now how it brought the game objects to the flywheel. A small strip of steel was placed at the top of the intake and ran towards the flywheel, creating a consistent path for the objects to follow. This final addition created a fairly consistent path for the game objects to follow towards the launcher. This design can be seen below in Figure 2. Figure 2 Final Large Robot Construction 6

11 Launching Mechanism Figure 3 Initial Launching Mechanism From the early stages of design, the launching mechanism of the large robot was narrowed down to a few different designs: a flywheel launcher, a linear puncher, or a target tracking turret using either a flywheel or linear puncher. A flywheel launcher was found to be the fastest implementable design, while potentially keeping a high consistency of shots. A prototype for this launcher was built, driving two 4-inch wheels with a chain-and-sprocket system creating a gear ratio of 1:16. Two of these systems were built, each driven with two DC motors geared together, each with 100 RPM. This gave an overall RPM of 1600, and a launch velocity of mph; the calculations for this launcher were completed using Equations 1 and 2, with the MATLAB code in Table 6 in Appendix A. This original prototype launcher can be seen in Figure 3 above. After building this prototype, a test of the range and height was conducted; the results from this test can be seen in Table 2. 7

12 Table 2 First Launcher Range and Distance Tests Trial Range (ft) Height (ft) Average Although the average range and height of the launcher was the exact requirements of the competition, a height of 3.5 ft and a distance of 18 ft, it was at a full power being set to the motors. With the potential of using PID control, Proportional-Integral-Derivative control, the launcher had to be redesigned to keep the motor s power setting at around 50-75% when launching that distance. This was needed because PID control checks the value of rotations at the motor, and, if they are under or over the required value, it will adjust the power setting of the motor to get it to the needed value [5]. With this knowledge, a new flywheel was designed to be 3D printed. Rather than keep the smaller 4 in diameter of the standard wheels used before, this new design had a 4.7 in outer diameter that tapered to a 4.6 in inner diameter with a height of 2.8 in. The initial CAD design can be seen below in Figure 4. 8

13 Figure 4 First CAD Design of Flywheel A new gearbox was designed to adjust to these newly designed flywheels. An internal motor gearing of 240 RPM was used with an external compounded gearbox of 1:9, giving an overall RPM of 2160, a launch velocity of mph, and an energy transfer of lbf. Calculations for this newly designed launcher were completed using Equations 1, 2, and 3, with the MATLAB code in Table 6 in Appendix A. v 0 req = x x 0 t cos θ (1) v 0 = r ω (2) TE = PE + KE = 1 2 kx Iω2 (3) Lift The decision to incorporate a lift into the large robot was decided a couple of weeks before the first competition. Before both robots were completed, the strategy for them was to clear most of the game field in autonomous, removing most of the opportunities for the other team to score. However, as 9

14 the competitions drew closer, this initial plan was looking like less of a reality, and a safety net had to be implemented as a last chance to turn around a losing game. Two initial designs were thought of for the lift. The first of these designs was an extendable ramp over the large robot, with a simple mechanism to release the ramps, and drop them. This design, however, was too bulky to fit within the 24 in cube size requirement and still have both the launching mechanism seated where it was and the rather large intake system. The second design was slightly more complicated, but had the entire lift within a 20 by 5 in profile that could mount vertically on the back of the robot. It called for a scissor lift that could deploy off the back of the large robot, allowing for the small robot to drive up, and the lift to activate and elevate the small robot. The decision to adopt a scissor lift was rather easy, as it was determined to be a relatively small profile that could mount to the back of the robot, and still fit within the size requirements. Scissor lifts can be constructed to lift in a number of ways; for instance, during the previous year, the team built a scissor lift with drive motors mounted on the bottom of the structure, driving the mobile legs along a linear slider with a rack and pinion system to provide the motion. This design proved to be horrendous for a scissor lift, requiring more torque to lift the system and bending some of the structure along the bottom. This lead to multiple rack gear failures during the year, unless the utmost care was taken to insure the lifting motion occurred slowly. Due to these past experiences, the team decided to build a scissor lift with more of a linear torque requirement. To accomplish this, stationary gears with 84 teeth were hard mounted, mounted so that as they moved, the channel they were mounted to would also move, to the legs of the lift, with a small pinion gear driving it, compounded to a 60 tooth gear driven by another pinion, which in turn was driven by a 36 tooth gear driven by a pinion attached to a motor geared internally to 100 RPM. In turn, this gave the lift a 105:1 gear ratio, more than enough torque to lift the near 8 lbs small robot. 10

15 The first iteration of the lift build mainly consisted of aluminum c-channels and small aluminum strips. After a single test run, the aluminum strips were found to bend and twist like previous designs, so they were replaced with an eighth-inch thick steel plate made from a linear slider guide rail, with the rails cut off. These were made because they were thicker and stronger than the standard small steel strips that were initially considered to replace the aluminum ones. During that same test run, extreme shaft deflection was seen with on the drive shafts other than the one driving the hard mounted gear. The easiest and quickest solution to fix this was to add bracing along the outer edges of the shafts, giving a support of each side of the gear. This reduced almost all of the shaft deflection and made the lift stable when lifting. A test run of the lift, before mounting it to the large robot, can be seen below in Figure 5. Figure 5 Lift with Small Robot On Top Mounting the lift, built at the end of the large robot build cycle, proved to be a challenging and time consuming task. The lift was built and designed separately from the large robot and no system to mount it was devised. The two sections where the small robot drove up ended up being spaced perfectly to the inside of the wheels of the large robot chassis and the support beams for the intake. Even though the lift fit there perfectly, an issue arose on how to deploy the lift, without breaking it. 11

16 A small standoff was mounted to the support beams for the intake, and a strip of aluminum was attached to the hard mounted gear of the lift and bent into a hook shape. The lift would then begin to raise while it was still mounted vertically to the chassis, and the hooks would unhinge from the standoff. Once they became unlatched, the robot would drive forward slightly, giving the lift a small amount of momentum to drop down, flipping the ramps for the small robot to drive up down with it. The final mounting solution of the lift can be seen below in Figure 6. Figure 6 Lift Mounted to Back of Robot Small Robot (First Iteration) Unlike the large robot, the small robot was initially designed to be a close-court shooter, where it launched the game objects from anywhere less than 12 ft from the goal. Drive System Early concepts of the small robot drive called for a hybrid setup, where the front wheels were omnidirectional wheels and the back wheels were high traction wheels. The intent of this design was to 12

17 allow the robot to pivot from the back when rotating. The drive was the fastest during testing, clocking an average of 8-seconds through the obstacle course; these drive tests can be seen in Table 3. However, the drive suffered from a lack of consistency in movement caused by slipping in the front omnidirectional wheels. Table 3 Small Robot Drive Comparisons Drive System Time A (s) Time B (s) Average Time (s) Tank Drive Total Average (s) Standard Deviation Hybrid Drive Total Average (s) Standard Deviation Percent Difference (%) Drive selection for this robot then followed the same logic as the large robot drive. A mecanum drive would offer the electrical engineering team the most control over the movement of the robot. Mecanum drives have minimal slippage, which means that the Integrated Motor Encoders (IMEs), a sensor that reads the rotations of each individual motor, creating a feedback system for the autonomous control, would have more accurate readings. This compromise did result in a slower drive, taking 17-seconds to navigate the obstacle course, meaning the drive is 212% slower than the hybrid drive. The benefits of having accurate readings from the IMEs was the deciding factor when selecting the mecanum drive. As mentioned earlier, the small robot needs to fit in a 15 in cube. Because of the small size factor, some alterations from a standard direct-driven drive had to be made. In the standard setup, the front motors would interfere with the intake. Therefore, the motors had to moved elsewhere, while the wheels needed to stay in the standard location. A chain-and-sprocket system was developed to allow the motors to be located above the wheels where there would be no interference with the intake. While there were 13

18 early concerns regarding slippage in the chain, thorough testing showed that any slippage by the chain was minimal and did not appear to affect the IME readings. Intake Device For the small robot, the intake proved to be the most challenging aspect. While the large robot intake leveraged large 1.3 in rubber flaps along a chain to guide game pieces into the launcher, the small robot could not accommodate such a large feature. The intake needed to be low-profile and maximize the amount of space available for the launcher, all while still capable of acquiring at least three game pieces. Early on the team decided that the intake needed to be focused on a simple chain. Using a basic chain would ensure that the intake could maintain a small size and still move game pieces, assuming there was enough pressure against the chain. The first prototypes suffered from game-affecting dead zones, or were incapable of possessing any more than two game pieces at a time. The most important aspect of the intake is the first-contact portion. This end, which sits at the front of the robot, contacts the game pieces first and needs to initiate control over as many as possible. The first prototypes of this section used the 1.3 in flaps wrapped around sprockets. These were ultimately too large and would violate the 15 in cube requirement. After thorough testing, four 1.5 in VEX intake rollers were shown to be the most effective. These rollers consist of twelve small rubber flaps that could consistently grip game pieces while not taking up too much space. Throughout development of the final intake, dead zones continued to crop up. Multiple solutions were implemented to reduce these dead zones, all of which focused on a metal ramp running the expanse of the horizontal portion guiding game pieces to the vertical portion. At first, string was implemented to provide pressure against the ramp, forcing game pieces against the chain. Unfortunately, the string loosened too frequently and would revert back to causing dead zones. The final design of the intake consisted of three primary sections: the first-contact, the horizontal section, and the vertical section. The first-contact, as stated, implemented rubber intake rollers to 14

19 maximize grip on the game pieces. After being gripped, the game pieces entered the horizontal portion. This section used a metal-reinforced Lexan ramp to provide light compression against the chain to help the chain pull game pieces through. After reaching the rear of the robot, game pieces entered the vertical section, which used a strong aluminum backing to provide the necessary pressure to elevate game pieces directly into the launcher. In this vertical portion, cut-to-fit sections of Lexan were placed along the track spaced roughly as wide as a game piece, 4 in, to ensure consistent feeding of the launcher. Launching Mechanism Similar to the large robot, the choices in launching mechanisms for the small robot were narrowed to a few concepts: a flywheel launcher, a linear puncher, and a pneumatically actuated catapult. Just with the large robot, the flywheel launcher was the cheapest and fastest method of building a launcher. Rather than use the dual flywheel design of the large robot, a single, horizontal axis of rotation was used for the small robot. The main reason for this single horizontal rotation flywheel was the need to have the entire design fit inside of the 15 in cube size constraints; the large robot flywheel was just over 18 in in length, with this new small robot design just over 14 in. This design created a longer path of contact for the game objects, decreasing the overall transfer of energy and launch velocity due to frictional losses. 15

20 Figure 7 Flywheel and Hood of Small Robot The small robot s flywheel uses a 240 RPM DC motor that drives a 1:15 compounded gear ratio, outputting a total RPM of Before accounting the frictional losses in the system, the launch velocity was found to be mph, transferring lbf of energy to the game objects. Calculations for the lossless launch rates were completed using Equations 1, 2, and 3, with the MATLAB code found in Table 6 in Appendix A. As seen in Figure 7 above, the small robot flywheel has a travel distance of almost 7-inches of travel from the exit of the intake to the launch point of the flywheels, decreasing the total transferred energy and launch velocity. This travel distance gives an average peak of 12 ft at a distance of 6-feet. Because the height of the goal is only 3.5 ft, the high arc allowed for the shot to drop into the goals from a distance of around 10 ft, without the chance of it being blocked by another team. CSM Skills Challenge Robots While the game mechanics are essentially the same, the skills challenge requires a completely different strategy to do well in it. Rather than having two teams compete against each other and have to share the field, each team is allowed to play the full field on their own for 60 seconds. This includes 64 16

21 driver loads that would normally be split between the two competing teams and the two goals that would normally be limited to 1 team or another. A concept for a relatively simple launching mechanism, that could drive straight across the field and play the driver loads, was proposed and built. Skills Challenge Robot (First Iteration) After performing at the competition in Purdue, the team acknowledged the need for a better and more consistent launching mechanism. After analyzing what the current issues were, what was causing them, and what other successful teams were using, the team decided to go with a cam launching system instead of the previous flywheel set up. Using a cam launching system provides a nearly perfectly consistent launching platform as well as the ease of compacting and fitting the robot within the dimensions for either of the two robots. This robot was designed around the main cam launching mechanism; due to this, no intake mechanism was needed, as it would only launch the driver loads on both sides of the field, and not go around picking anything up off the field. Drive System The skills robot focused entirely on firing the driver loads, meaning it only needed to stay on one side of the field rather than playing the field. This means that the skills robot did not need to have a high mobility drive system like the previous robots designed. With this in mind, standard 4 in high traction wheels were used. These wheels offered optimal precision when navigating from one end of the field to the other, which helped since it only needed to move in a single, horizontal line across the field. Using the high-traction wheels over the mecanum wheels offered optimal speed while maintaining control. Launching Mechanism Two designs were considered for actuating the liner punching mechanism. The first idea was to fabricate a slip gear by removing 6 sequential teeth from a 36 tooth gear. This gear would mesh with a rack set, attached to a hammer, and would pull the hammer back and release it. The second idea was to create a logarithmic spiral shape out of 1/16 in polycarbonate to act as a cam to pull back a shaft attached to the 17

22 hammer. Through group deliberation, it was determined that a spiral cam would be designed, cut, and tested for reliability, while the slip gear idea would be kept around as a backup plan. Figure 8 Spiral Cam Equations A 3D model of a logarithmic spiral cam was created using the equation curve function in Autodesk Inventor, this can be seen in Figure 8 above. The spiral shape itself was designed using the following parametric equations [6], these equations can be seen below. { x(t) = r(t) + cos(t) = aebt cos(t) y(t) = r(t) + sin(t) = ae bt sin(t) (4) This design has gone through four revisions. All four revisions can be seen below in Figure 9. The original cam was cut out of polycarbonate. The spiral shape was created using the spiral tool in CorelDRAW by Corel. This cam has an inconsistent shape, and would cause the hammer to get caught up on the backwards pull. The black cam, which was the first cam designed using the equation curve, only had about 2.5 in of travel. The short travel on this cam would theoretically allow for an extremely fast launcher with minimal torque required. Through testing it was found that more travel was required. More travel allowed for less rubber bands to be used, reducing the risk of damage to the linear sliders and hammer. To keep 18

23 the part within the 6inx6in size requirement, the inner circle of the spiral was tightened, and the outer circle diameter was increased slightly. After this revision, the travel distance was improved from just under 2 in to around 3 in. This small change greatly improved the overall actuation of the hammer. Figure 9 Four Revisions of Cam The movement of a cam launcher is based on two parts: the sliding hammer, and the rotating cams. In order for the hammer to move in a consistent and stable linear motion, two rails were installed onto a test frame. Each of these rails was mounted at two points that were spaced out far enough to allow for possible changes in the pull back and launching motion. The hammer had guides placed onto them to allow them to follow and track the motion of the cams. There was then a padded stopper installed onto each rail to stop the hammer before their guides would ram into the cams when in the launching motion. To power the system, two motors, with 1.7 Nm of torque, each were geared to 33.3 RPM to rotate the cams; and four to five rubber bands were mounted on each side of the hammer going from the cam guides to the test chassis. 19

24 Preliminary tests with the first prototype launcher found the new system to be a vast improvement over the original flywheel systems the team had been using. The cam launcher was placed approximately 7 ft away from a goal, aimed at roughly a 60-degree angle, and had four rubber bands mounted on each side. With this configuration, the launcher was able to fire four shots in roughly five seconds, and would make about 90% of its shots into the goal with the primary reason for misses being inconsistent or bad loading of the game pieces. With the knowledge gained from the early performance tests and prototype launcher, a few changes were considered to improve the performance and accuracy of the launcher. The first change made was the addition of a 3D printed ball holder placed at the end of the launcher that held the game pieces until the hammer makes contact with them. This helped standardize and simplify the loading process, which helped increase the speed and accuracy of the launcher. The rails used for the hammer were oiled from end to end to allow for a smoother launching motion. These changes and modifications can be seen in the completed robot below in Figure

25 Figure 10 Cam Launcher Mounted to Support Structure Large Robot Changes (Second Iteration) The large robot, in order to be more competitive in the skills challenges, needed to be redesigned, especially the intake device so the robot could consistently pick up a full stack of game objects. Drive System The drive for large robot redesign was almost identical to the previous iteration. The only difference between the two was the mecanum wheels were switched for omni direction wheels. The omni direction wheels are much faster than mecanum wheels, and allow the robot to turn easier. Intake Device An intake for the large robot had to be designed specifically for the autonomous skills challenges. The previous version of the intake was only able to pick up one or two of the game objects in a stack consistently; to post competitive scores, all four game objects from a stack needed to be picked up every time. 21

26 Throughout the initial concept meetings for this new intake device, various designs were proposed; the best concept developed was to make the intake mounted to the chassis shorter, with a front gate mechanism mounted on the front of the chassis that swung down to entrap the game objects in a stack inside the intake. The original chain and sprocket sections of the intake held the same purpose for the newly built device, but had a significantly lower profile. The intake was built in the shape of an ellipse with the minor axis set to the height of two game objects. The lower profile allowed the intake to be smaller, which reduced the distance the ball had to travel to the launcher. The flaps on the top of the ellipse brought the game object directly to the launcher, helping with consistency issues of launching. The changes made to the large robot from Purdue can be seen below in Figure 11. Figure 11 Intake Changes made to Large Robot 22

27 World Championships Robots The team will be attending the 2016 VEX U World Championship in which 60 of the 300 currently active teams registered in the world coming from six continents and twelve different countries are competing. ODU will be bringing three robots, where two of them are specialized for the head to head competition, and the final will be used exclusively for the skills challenge. Large Robot (Third Iteration) The large robot, much like the previous designs used at Purdue and the CSM Skills Challenge, needed to be able to score both consistently during autonomous, as well as score all 32 of the driver loads it was allocated during a tournament match. To accomplish this, multiple designs were proposed based on past experiences during the building process and strategies seen at competitions. The two contenders for consideration were a robot with a dual launching system: a flywheel on the left half of the chassis, and a cam launcher on the right half, or a robot that was back-fed to easily accommodate the operators for driver loads, while still being fast enough to drive around the field and either score points or employ a defensive strategy during a match. After much talk and deliberation, the decision to go with the second design was chosen. Drive System The drive system on the world s large bot was similar in design to the old robot. Four omnidirectional wheels were used in a square design. The wheels were an equal distance apart in the x and y directions. One of the issues with the CSM skills bot was that the center of rotation, and center of gravity were in different locations. This caused the robot to spin unevenly, which made the autonomous movement inaccurate. This issue was solved in the new large bot by mounting the weight evenly. The new large bot was able to spin much more accurately. Intake Device The world s large bot intake design was very different than the Purdue and CSM large bot. Instead of intaking through the front, and shooting out the front, the new large bot intakes through the back and shoots through the front. This design allowed for a much lower profile, because the game objects could 23

28 be fed in a straight long instead of looping up and around into the launcher. For perspective, the base of the old launcher sat at 16 in, while the base of the new launcher sits at 7 inches. This design lowered the total distance the game objects had to travel. The new intake was split into two sections, similar to the CSM large skills bot. The first section comprised of two sprockets spaced 15 in apart, spinning at 100 RPM. Surgical tubing is held under tension, between the sprockets. This intake is very effective at picking up balls because the tubing grips the game objects as it moves through. This intake is able to pick up to three game objects reliably. The second section of the intake is a flap and chain system, which is exactly the same as the previous iterations of the large bot. Launching Mechanism As with the drive, the flywheels were pulled from the large robot taken to Purdue and CSM and modified. The main gearbox and motors were kept the same, while the wheels and orientation of the launcher changed. The universal joint originally used for the launcher was removed, and the wheels were placed on the same shaft as the final gear in the gearbox. Additionally, standard 4 in VEX wheels were added to the base of each flywheel as added weight, to increase the momentum on the flywheels, and thus decrease the spool up time required between each shot to make accurate ones. The entire system was then mounted to the chassis at a 45-degree angle, so it could make high goals from anywhere on the field, even next to the pipe at the low goal. Figure 12 below shows the changes made. 24

29 Figure 12 Flywheel Changes for Worlds Large Robot Lift A new lifting mechanism was designed for the world s large bot. It was determined that a sliding ramp system would be the most effective lift for the new large bot. Due to the lowered profile, a ramp could easily fit on the top, and stay within the size constraints. After testing, it was determined the steepest angle the new small bot could drive up was 20 degrees. In order to get the bonus points, the ramp had to be at least 12 in tall. The ramp was mounted on an axle, so that when the small bot drives to the top, gravity caused the entire lift to rotate from 20 to 0 degrees keeping the small robot above 12 in. Sliding and hinge mechanisms were designed to keep the ramp within the 24 in cube. The end of the ramp was mounted on hinges, and folded back into the robot. To deploy the ramp, 2 pneumatic pistons push the hinges, and the end of the ramp folds down. The next section of the lift was mounted on sliders, with a rack gear spanning the length the c-channel. A motor spins a gear which then slides the channel down to the desired length. Figure 13 below shows this new lift. 25

30 Figure 13 Lift Mounted on Worlds Large Robot Small Robot (Second Iteration) The small robot, just like the one built for Purdue, was designed to be an extremely fast and mobile robot, however, improvements were made to both the intake and launcher to pick up a complete stack of game objects and launch faster. Drive System While the original drive system on the small robot was built, it was intended to be an extremely fast and mobile system to get around the field as fast as possible and pick up as many pile of game objects as it could. It was, however, slowed down and torqued up to suit the electrical engineering team s needs. The newest design was made to suit the original purpose of the small robot, mainly due to the improvements made to the control code by the electrical engineering team. The drive was designed with four motors, each rotating at 100 RPM, driving a gearbox to gear up the speed 1:3 on 3.25 in omnidirectional wheels, giving a maximum possible speed of 2.9 mph. Figure14 shows the initial construction of the chassis. 26

31 Figure 14 Drive System for Worlds Small Robot Intake Device After the chassis was constructed, the initial concept for the intake was prototyped: a passive bar that flips up as it drives over a stack of game objects, and sucks them in as it drives past them. This was built to feed the game objects into an inner section that feeds the balls directly to the flywheel launcher. Figure 15 below shows this bar on the finished chassis. 27

32 Figure 15 Passive Intake on Front of Chassis In order to build the rest of the intake, the flywheel had to be built and mounted; a vertical intake section was then built to raise the game objects from the floor and bottom section of the robot to the top. It was driven by two motors, each attached to a bevel gear to turn the shafts 90-degrees from the motors mounted orientation. The bevel gears drove a shaft that had a sprocket mounted on the end of it, which drives tank-treads with flaps mounted to it. Those flaps lift the game objects up about 10 in, and pushes them into the flywheel. Figure16 shows the top section of this intake. 28

33 Figure 16 Hood on Worlds Small Robot Launching Mechanism After seeing the inconsistencies of the small robot from Purdue, new designs for a launching mechanism were proposed: a pneumatically activated catapult arm or a redesigned flywheel launcher, built off of the previous one. While a pneumatically activated catapult would be more consistent and potentially score a larger amount of points, it was seen as too difficult to build in the short time frame of two weeks before the World Championships were held. Because of this, a new flywheel design was chosen as the go-to option. The first change made to this system was a new 3D printed flywheel: a 5 in diameter, 1.5 in high cylinder was designed in Inventor. Unlike the previous flywheel, this new design was made to be printed at a high infill, to add weight to the piece. Additionally, two of the standard 4-inch diameter VEX wheels were placed on either end of the 3D printed wheel to add even more weight to the system. The previous flywheel, shown in Figure 7 weighed only 0.26 lbs, while the new design is about 0.77 lbs, nearly a 200% 29

34 increase. The added weight increases the momentum of the flywheel, which, in-turn, decreases the amount of spool time between each shot it can make, meaning the robot can launch game objects faster than before. Figure 17 below shows the finalized mounting of the flywheel, with rubber bands wrapped around the 3D printed wheels to add traction. Figure 17 Flywheel Mechanism for Worlds Small Robot Due to the increased space requirements from the additional wheels in the center of the system, the compounded gearbox on either side had to be rebuilt and shrunk down. The overall gear ratio was kept the same, 1:15 driven by 240 RPM motors, but the size of the gearbox was brought down from 2.5 in wide to 1.5 in wide on either size, plus the additional space for the two motors driving the gears. Additionally, the hood, a metal plate that compresses the game objects as it goes to the flywheel, was changed, decreasing the overall compression of the object, as well as the total contact time between it and the wheel. 30

35 All these changes made on overall significant impact on the performance and consistency of the flywheel system. The shots it makes are aligned with the goals nearly 95% of the time, an increase from the previous 80% of the time. The launch speed changed from mph to mph; this stems from the lower compression and contact time of the flywheel. Calculations for the linear launch velocity was done using Equations 1, 2, and 3, with a MATLAB script that can be seen in Table 6 in Appendix A. Skills Challenge Robot Changes (Second Iteration) The skills challenge robot, while highly consistent and accurate, had a few issues that needed dealing with before it could compete at another competition: a drive issue and a loading inconsistency. Chassis Modifications The robot only needed to drive across the field in a straight line and hit the wall on the other side, however, as it drove across, a stack of game objects located on the field would interfere with it, and potentially cause the robot to lodge a game piece between it and the wall, rotating itself away from the goals. To counter this, a set of rails were mounted to the bottom of the chassis, with a plate that was curved to push the game objects away from the center of the chassis. As the robot hits the game field perimeter, the bar gets pushed in, hitting a limit switch on the inside of the slides, stopping the robot so it can launch game objects. Launcher Additions The launcher had user error loading inconsistencies. The problem was that the user would load the game objects into the holder too hard, forcing the ball into the holder, instead of allowing the ball to rest on the holder s chamfer. The added resistance of the compressed ball pushing against the inside of the ball holder would lead to a lower launching velocity, thus the balls would fall short of their target. There was also a potential of bending and eventually breaking the holder, since it was a relatively thin part. To alleviate these issues, the of the opening of the holder was decreased around a quarter of an inch, making it so the operators could not force the game objects in; the overall thickness of the holder 31

36 was increased by around a quarter of an inch, so that it could not be bent and broken in half. With this increased thickness, countersunk screw holes were turned into countersunk slots to allow for more adjustments of the location of the holder. Figure 18 shows the old ball holder on the left, and the new ball holder on the right. Figure 18 Old (left) Ball Holder Compared to New (right) Robot Programming During the early design phases, autonomous programming was a central part of the team strategy. The coding is split in to two main categories: the autonomous and driver control code. The code 32

37 for the robots can be found in Table 7 (Large Robot), Table 8 (Small Robot), and Table 9 (Skills Challenge Robot). Autonomous The autonomous section of the design employed a field centric perspective rather than a more common robot centric view. Rather than have the robot search for game objects relative to itself, the static field objectives were inserted onto a virtual game field that the robot moves around. This virtual field allowed the programmers to discretize all autonomous robotic movement to ensure that proper placement and motor movement occurred at the proper times. With this setup, rather than specifying any wheel movement in any direction, the microcontroller essentially sends a signal to the robot telling it to move to a specific (x, y) coordinate on a virtual field. Since the tiling scheme is discretized, there is a specified angle required to score at any point in the field that is standardized and accurate regardless of the robot s current tile. The autonomous procedure tracks the position of the robot on the virtual field in terms of tiles; the game field is six tiles wide and six tiles long. Since balls aren t positioned directly centered on tiles, half tiles are also supported. Motor encoders ended up being the ideal method for tracking movement, as accelerometers and gyroscopes were either too inaccurate or experienced significant drift over the course of the match. Using the physical characteristics of the wheel, the programmers were able to determine the exact number of encoder ticks required to travel a single tile in any direction, as well as 90- degree rotation. Functions were created that mapped tile coordinates to the required number of encoder ticks from any given position. The launching mechanism is a variable speed flywheel that is modulated depending on the distance to the objective. Due to the field setup, the robots are able to know a relatively accurate distance to the objective at all times. Four discrete motor speeds sufficient for the different distances, from full field shots to shots just squares away. Depending on the coordinate of a shot, the flywheel speed is 33

38 modulated to the specified distance setting and maintains that speed for the duration of the shooting period. One concept that has been employed to obtain more consistent launching without constantly monitoring battery power is a PID like system that relies on the motor s RPM rather than power levels. Each robot was calibrated for specific shooting distances and recorded the RPMs associated with those positions. When shooting in both autonomous and driver controlled sections, the microcontroller constantly monitors the RPM of the motor associated with its current speed. If the RPM falls short, the intake is paused and power is increased until it reaches the threshold RPM. This not only causes consistent shooting, but also allows the robots to fire more balls in a row without having to manually pause the intake. The autonomous procedure consists of moving the robot to the specified square, intaking game objectives, rotating to the objective and launching, and rotating back. Since objectives are static and stored on the virtual game field, the code can simply specify which objects the team wishes to pick up during the autonomous section using constants. Each objective will be picked up and fired in the order in which they are entered. Since there are two starting areas to the field, normally it would be required to create two separate autonomous routines in order to get full game coverage. Since a virtual field is created, the only thing that the programmers need to do is change the starting coordinate, and the program will handle the rest of the field and object conversions required to complete the routine. Driver Control The driver control code can be separated into three different sections, intake control, position control, and launcher control. All of these tasks are run concurrently using a task system. The user is free to adjust position, rotation, intake speed, and launcher speed without any fear of interference. The driver control section of the code design reuses the launcher principles of the autonomous section, while transferring all control to the user. The controller s left side buttons were mapped to four 34

39 different speeds, descending in clockwise order. After user feedback, the two fastest settings were also mapped directly to the right side triggers. When one of these buttons is held down, the launcher speed is modified to the speed specified by the button and held there until released. A base level of constant speed is used to reduce the time between a button being pressed and the launcher reaching the desired speed. The same PID like concept employed in the autonomous section is used here as well. Each preset flywheel speed has a RPM associated with it, and the intake is paused while the RPM is not at its threshold. The intake controls are extremely simple, and can be reduced to the use of two buttons. The left side triggers are used to control the direction and operation of the intake, where the upper trigger feeds the balls towards the launcher and the lower trigger pushes them outwards. The outwards button was included to help combat intake jamming; whenever a ball lands itself in an unfavorable position, the user is able to back out the intake, unlodging the ball. The position control is a simple joystick mapping directly to the wheels. The joystick is analog based and therefore supports proportional output, giving the users finer control over both the speed and direction of the system. The left stick is in control of the two axis movement throughout the field, while the right stick controls positive or negative rotation. These controls were split into two channels to prevent any accidental movement, as well as to aid in the aiming of the robot during launching phases of user controlled operation. 35

40 Volunteering Along with the competition and robot design, volunteering and community outreach is an extremely important aspect of the team. Throughout the year, ODU was able to volunteer at regional conferences and VEX competitions, local schools for STEM nights, and local businesses and attractions. High School VRC Events Figure 19 Andrew and Allan Inspecting Robots at the Hampton Roads VRC Event The ODU VEX U team has a strong partnership with Portsmouth Public Schools to help mentor teams, as well as volunteer at robotics competitions they host. The day after ODU s first competition in the Skyrise season, the entire team volunteered at the 2015 Hampton Roads Spring Regional VRC Event. They served as referees, inspectors, and the emcee. This year, they asked the team to come back for both the 2015 Hampton Roads Fall Regional and 2016 Hampton Roads Spring Regional for the Nothing but Net season. Due to the reduced number of referees this year, the team served as the head referee and supporting referees, judges, inspectors, the emcee, and the photographer. These events served as a great experience for the team, since members are able to have an impact on the next generation of scientists and engineers. 36

41 Figure 20 Richard and Mike Serving as the Head Referee and Emcee Larchmont Elementary School STEM Night Figure 21 Richard Volunteering at the Larchmont Elementary School STEM Night The team was invited by their project management professor to take part in the STEM Night at Larchmont Elementary School in Norfolk, Virginia. They showcased to the next generation of scientists and engineers what you can achieve when great young minds work together to conquer their goals. Unfortunately, only the small robot was fully function before the event, but they were able to showcase the autonomous code for the large robot chassis, as well as a mechanical gyroscope that was accidentally developed while working on the large robot flywheel. 37

42 Both the students and parents got involved, asking questions about all of the systems the team had on display, especially the 3D printed flywheel and logarithmic spiral cams, as well as the mecanum wheels and how they worked. The event was a huge success by the team s margins, as they taught a very young group of students about robotics, something we hope sticks with them as they finish school, and potentially pursue STEM careers and 2016 Old Dominion University Admitted Students Day Towards the end of the both the 2015 and 2016 spring semesters, the team was asked to participate and set up an informational table at the annual ODU Admitted Students Day on campus. At this event, potential freshman came to the school to learn about the campus, majors, dorms, and current students to determine if they want to attend ODU after they graduate high school. The team served as ambassadors and helped possible freshman learn about living on campus as well as what it is like to be an engineering student. Freshman ENGN 110 Last semester, the team partnered with their faculty advisors, Dr. Chen and Dr. Hou, to develop lesson plans and teach freshman students about robotics design, mechatronics, mechanical engineering, and electrical engineering in the Introduction to Engineering course at ODU. Michael presented the mechanical engineering portion of the lectures, delving into robotic design and how the mechanical team goes about analyzing the robotic systems, while Daquan presented the electrical engineering portion of the lectures, discussing the power systems, sensor integration, and code development for the robots. An interactive presentation was developed to keep the students engaged and interested, while also giving them assignments to receive a grade. To increase interest in the lectures, YouTube videos and robot demonstrations were utilized. A functional robot was assembled based on a number of prototype systems, including a chassis, intake, and launcher. 38

43 During the class, a heavy interest was developed among the freshman engineering students in the organization and robotic design; that interest will hopefully guide the students in their future career choices. Homecoming Parade For this year s homecoming parade, we partnered with the National Society of Black Engineers, the Society of Hispanic Professional Engineers, and the American Society of Mechanical Engineers to design a float to show off the engineering organizations on campus. While the other organizations focused on building the float on a trailer for the parade, the ODU VEX U organization built a robotic catapult to launch candy into the crowd using surgical tubing to get the distance needed. Parts were sourced from ServoCity and SparkFun so that the robot built could be kept for a number of years. The entire system was controlled using an Arduino UNO, with a control system run using switches and buttons. The catapult was charged using a slow, high torque DC motor powered by a 12V lead acid battery and surgical tubing. As the centerpiece of the float, the robot had to launch about 15 feet in both a safe and consistent manner so as not to harm spectators, which was able to be done by calibrating the surgical tubing. Figure 22 Homecoming Robot on top of the join Engineering Float 39