Mechanical

Accomplishments:

The four major components we tackled this week were the clutch, the drive for the arm, the camera for remote viewing, and the reloading mechanism.

Clutch

The clutch mechanism was switched from an electromagnetic clutch to a mechanical-only dog clutch on a custom-made 1/2" steel shaft. The mechanical dog, between the red bearing block and the black belt, rides on the hex portion of the clutch shaft. It is pinned through the clutch shaft, which connects it to the shifter block on the right of the black pulley. By moving the shifter block, the dog can be engaged or disengaged from driving the black pulley. When disengaged, the black pulley spins freely on the shaft and is captured by an e-clip. Two bearings capture the shaft and resist the forces from belt tension. The two bearings are now in a sturdier unified holder, as the two individual holders had significant deformation under belt tension. This caused the belt to slip and the catapult to launch prematurely.

Motor

Once the new clutch was put in, we faced a number of problems in actually getting the motor to pull down the arm:

• Flex coupling between 3D-printed motor coupler (so printed due to the double-square spline necessary to interface with the window motor) was slipping on the clutch shaft (shown in the above image). As the single-sided split coupling mechanism of the flex coupler could not be tightened further, the flex coupler was pinned through the shaft.
  • Testing impossible due to excess slippage.
• To prevent further slippage of the flex coupler, the motor coupler was CNC machined and pinned through the flex coupler. The motor mount was also switched from 6mm birch ply to 1/4" alumnium.
  • Upon testing, the motor reached stall when the arm reached 45 degrees from horizontal; this is a far cry from the parallel to horizontal that we need to obtain all the force from our six-spring setup.
• To increase the motor torque, we scavenged a 10:1 gearbox from an industrial-grade 3-phase 160VDC servo motor. A new motor coupler was CNC machined to be a running fit on the gearbox input and the flex coupler was bored out to accept the gearbox output. In place of the 3mm Woodruff key required on the gearbox output, a turned-down set screw is used to fill the key slot. The gearbox was mounted to the frame via a 3D-printed two-piece mount.
  • Upon testing, the motor did not stall under any condition; conservatively, our motor is now geared down between 100:1 and 800:1. This increase in torque caused belt slippage due to insufficient wrap on the smaller 10T pulley. This lack of wrap is due to the large size difference and close proximity of the 10T and 40T pulley. The two red and yellow bearing mounts, shown in the photo above, were also deforming drastically under the belt tension.
• Red and yellow bearing mounts replaced with a single double-ended mount. While still deformable under our expected loads, the deformation is small enough to be taken up by a belt tensioner. A belt tensioner was added as an attempt to prevent belt slippage:
  • V1: Two spring-loaded L brackets with a bearing between them to tension lower track of belt. As the two bolts were perpendicular to the belt, the tensioner simply folded over under load.
  • V2: Custom printed assembly: Moved bolts parallel to belt and switched to shoulder bolts in reamed holes for smoother motion. Worked in principle, but the plastic base would deform under load. Belt rides on three small bearings attached to a shoulder bolt for smooth motion.
  • V3: Plastic base replaced with metal base. Base did not deform, but springs were insufficiently stiff and bottomed out with belt tension.
  • V4: Metal base drilled out to accept thinner shoulder bolts, for which we scavenged much stiffer screws. Springs were stiff enough to not bottom out, but this caused the other side of the belt to go slack and skipping behavior resumed.
  • V5: Realized that we were tensioning the tension side of the belt, not the un-tensioned side. Switched tensioner to untensioned side to good results, but tensioner obstructed other side of belt.
  • V6: Added back shoulder bolts and required clearance for other side of belt. Works well, moving the slippage problem down the line to the large pulley.
• The next step is griding flats into the main 1/2" shaft for the large pulley's set screws to capture.

Camera

Given the challenges from the camera we were originally using, we decided to move to a more robust and reliable platform. One of our team members had a spare phone, so we decided to connect to that through a wifi video chat service. We will use either Facebook Chat or Google Hangouts - one is higher quality but the other has less lag. Our big concern is that the final competition will be in a very busy building during dead week, so the wifi bandwidth may be severely limited. We plan to test it the week before the competition in case we need to find a backup plan.

Reloading Mechanism

The reload mechanism is fully assembled. There are two cups that sit on either side of the arm. They are controlled with individual servos, and will be turned one at a time.

Challenges:

The main challenge we keep facing that has affected multiple aspects of our design is how much torque is required for this game. Our system is able to throw the full 3m required, but in the process we have had to change and redesign multiple parts due to the forces required to hold the arm in place. As well, disengaging the clutch has been very challenging given how much pressure is on it, and we are still working toward a reliable mechanical solution for this.

Originally, we were using a pan/tilt camera that we had left over from another project. However, that camera had no documentation online. We were finally able to connect to it, but it provided a very poor quality image and had a very serious (3-5 second) lag.

Next Week:

The goal in the next week is to finalize the design and assembly of the clutch actuation mechanism, test the throwing mechanism, and begin integration of the full system. As stated above, the clutch mechanism is very challenging, so it may take longer than expected to come to a final design.

Electrical / Software

This week, we accomplished three major goals. The first was to re-spec, redesign, and assemble the clutch mechanism (minus a few final pieces that have not been delivered yet). The second was to integrate the radio controller with the Roomba for teleoperation. The third was to finalize the design and build the initial prototype of the loading mechanism.

Mechanical

Accomplishments:

The two major mechanical components we tackled this week were the clutch and reload mechanisms.

The clutch is in the final stages of assembly; we are still waiting on a few parts to be delivered. Our initial electromagnetic clutch was not able to hold the torque required by our design, so we switched to a much stronger dog-tooth clutch. The downside is that the dog-tooth clutch requires a separate actuator to engage / disengage. We tested two different release mechanisms, an electric and a pneumatic solenoid. We decided on the pneumatic solenoid for its power density and quick actuation. This, however, adds both a small compressor and a separate valve. We had initially shied away from this because the compressor is very noisy and adds a lot of vibration, but the other design we tried did not pan out. Regardless, we must machine a small but complicated partially-hex shaft whose design is still being worked out.

The reload mechanism will be mounted to two 80-20 aluminum posts on the sides of the catapult. Each will hold a wood cup, which will be above the launch arm when it is fully pulled back. The wood cups will be attached to a servo, which will rotate when that bean bag is needed. This week, we finalized the design and made the initial build of this system.

Challenges:

The clutch provided us many challenges along the way - our initial spec and the clutch we purchased were not able to hold the amount of torque we needed. This caused a re-design and second round of ordering for parts. We are waiting on the final parts to be delivered so we can assemble and test the full throwing mechanism.

One of our biggest challenges this week has been the limited bandwidth of all team members. Week 5 is always a tough week between project deadlines, midterms, and research deadlines. We have, however, found time to balance the workload and make incremental progress.

The camera, which we were told would be plug-and-play, is from China and has zero documentation available anywhere. We have tried connecting it to the campus visitor wifi, setting up a separate router, and connecting it via ethernet. We have yet to be successful.

Next Week:

Our major goal for next week is to begin integrating the individual pieces into a full system. The first step will be to unify the mechanical pieces into one system.

Software / Electrical

Our goal for week 4 was to complete an initial basic build of each major component of the system. This initial pass allowed us to see what of our design works, what details are missing, and which parts would not work for the system.

Mechanical

Accomplishments:

This week, we completed the initial assembly of the basic launch mechanism. We initially tested the system without the motor and found that it can launch the beanbag the required distance (approximately 10').

Challenges:

During assembly, we found a few issues with the parts we ordered. First, we currently have just enough torque to launch the beanbag the right distance, but only when the arm is fully extended. We would like to add additional springs so we have more room for error. This will allow us to refine the distance thrown by the amount the arm is pulled back and the launch angle. Second, we found that the clutch ordered is not able to hold the amount of torque required. Both of these issues will require additional parts to be ordered.

Next week:

By next week, we would like to have a fully functional catapult, which will then be mounted on the Roomba. Our goal is to have a functional system (aside from the loading mechanism), which can then be refined.

Electrical / Software

Our goal for week 3 was to advance the design to a state sufficient to place orders to build a full, albeit tethered (i.e. wall supplies instead of batteries) first pass design. This week, we finished a full mechanical design of the throwing arm prototype, sufficient to place parts orders and begin fabrication. Effort was made to use scavenged parts wherever possible so as to stretch our budget as far as it would go. Additionally, the electronics developed a control scheme using a hobby radio controller and the controls developed the code necessary for Roomba control via the radio controller.

Mechanical

Our mechanical design for the thrower is a spring-loaded torsion catapult.

The arm is attached to a live axle via a 40-tooth L-series timing belt pulley, which is press fit on the axle. This axle also has several large torsion springs, the quantity of which we will adjust as-needed, to provide the spring tension force. The live axle is captured by a pair of 1/2" ID pillow block bearings. The 40T timing-belt pulley is driven by a smaller 10T timing-belt pulley, which connects to the window motor via an electronic clutch. The electronic clutch is a 24V model of unknown provenance and maximum holding torque; these unknowns are the rationale for gearing down the window motor, as it will reduce the holding torque required of the clutch. When the clutch is released, the torsion springs will attempt to return to rest, and thus throw the arm without requiring the window motor to be backdriven. As the window motor used a worm drive, backdriving it is impossible.

Accomplishments:

• Lots of scrounging for parts / getting "creative".
• Updated CAD design (as shown on left)
• Orders out to McMaster (springs, timing belt) and Surplus Center (electronic clutch, pillow blocks)
• Make drawings for machined parts & begin machining

Challenges:

• Unknowns: feasibility of mechanism as designed (maximum clutch holding torque, clutch power consumption, spring constant needed, feasibility of making window motor shaft coupler)

Next week's goals:

• Full prototype of throwing mechanism
• Begin work on reload mechanism

Electrical / Software

Our electrical design strives for simplicity. The robot will be controlled via a R/C hobby controller. The R/C signals are interpreted by the Arduino into serial commands for the iRobot Create 2. The window motor and electronic clutch are controlled through a switchable MOSFET board. In later weeks, the electrical and software sections may be broken into separate sections, but they are combined here due to their tightly-intertwined nature at this stage.

Accomplishments:

• Parts selection for basic design (Arduino, MOSFET board, etc.)
• Successful scavenging of radio controller and ordering of radio receiver
• Had the Roomba talking and whirring (briefly)

Challenges:

• Accidentally killed the power supply for a knock-off Arduino when it was connected to USB and Vin at the same time. They are supposed to have a protection MOSFET for this, but apparently not.
• Roomba seems to not want to talk to us anymore, but it's not limited to the initial Roombas. We may have offended the entire species.

Next week's goals:

• Repeatable Roomba movement
• Integration of radio controller

Accomplishments

• Formulated team
• Division of responsibilities
• Initial design, logistics, schedule
• Find parts that are available, figure out gaps
• Initial group presentation
• Updated CAD design

Challenges

• One group member gone for week 2

Next week's goals