Waterjet Cutting and Actuator Simulation

Individual Progress

Machining

After cutting parts out on the waterjet (Figure 1), I:

• drilled and tapped holes in the bearing holders and bushing holders
• drilled and reamed shaft holes in the knee joint and motor mounts
• drilled and reamed bushing holes in the bushing holders, ballnut plates, and spring module bottom & top plates
• bored bearing holes in the bearing holders
• milled part surfaces as necessary to adjust thickness of parts
• sanded and prepared the parts for assembly

All of these operations were carried out in the RI Machine Shop.

Figure 1: Actuator stack components, cut using waterjet

At its present state, the actuator stacks for knee and ankle are complete. The knee joint is also complete (Figure 2). I still need to make the shank, ankle bearing holder, knee bearing holder, ankle joint bar, knee actuator distal shaft mount, and ankle actuator proximal shaft mount. I am considering making the shank with square tubing instead of round because fastening to it is much easier, even though square cross sections do not resist bending as well as round ones.

Figure 2: Knee joint and knee actuator installed in iWalk knee brace

Shank Moment Simulation

In order to determine whether square tubing will be sufficiently strong under bending, I am simulating the forces and moments on the current leg design using positions and forces measured during non-amputee human walking (Figure 3). I read the data from a spreadsheet of data obtained from Nitish Thatte in Professor Geyer's group. Using this data, I calculate the length of and force in each actuator over time. Using this information, I will calculate the moment on the shaft, and thereby determine the required bending strength of the shank.

Figure 3: Conversion between required joint torque/position and actuator force/length

Challenges

Taper of Waterjet Cut

Parts that are cut on the waterjet have an approximately 1 degree taper on the cut edges. This makes clamping and aligning the parts difficult. This time, we ended up sanding the parts flat on the disk and belt sanders in the RI shop. For positioning, we placed conical edgefinders in waterjet-cut holes and moved the mill table until the edgefinder was true. From there, we extrapolated to the zero point that we wanted based on the dimensions of the hole pattern. Next time, I will do the waterjet cut step last. This will allow clamping of the stock plate in the CNC mill or regular mill, providing easy positioning and clamping. The tricky part will be aligning the plate in the waterjet and zeroing the cutting head. However, I think this extra difficulty will be outweighed by the time saved sanding and positioning parts.

Leg Redesign and Machining Preparation

Individual Progress

Leg Design Improvements

After consulting with Ben Brown, I moved the series elastic actuator attachment points to be in line with the center of the actuator (Figure 1). This nearly eliminates bending moments on the actuator. Bending moments are bad because they increase the friction on the linear slide, potentially reducing the accuracy of the force measurement.

To accomplish this, I offset the shank, using the shank and the end attachments to bear the bending moment. This redesign also completely eliminated the previous 80/20 aluminum shank material. As a result of these design changes, the leg reduced in weight by about 0.75 kg.

Figure 1: Leg redesign for reduced mass

Machining Preparation

Most of the parts in the series elastic actuators in the leg have only 2D or 2.5D geometry, making them well-suited for cutting on the waterjet at TechShop. Therefore, I have been testing different speed vs. accuracy tradeoffs on the waterjet. I will have to ream out bearing and bushing holes, as well as the holes to mount the linear guide shafts. I will cut these slightly undersized on the waterjet. I confirmed that screw tap and clearance holes come out accurately.

Finally, I took the RI Machine Shop safety training from Chuck, so I can now access the RI shop. While I plan to cut the outline of the parts and cut the tap and clearance holes on the waterjet at TechShop, I will try to do milling and CNC milling at the RI shop. The CNC milling may not be necessary for most of the parts now that the design has been simplified.

Purchasing

I selected appropriate flat stock (7075 aluminum, 12”x12”x0.5”) and ordered it. In addition, I found appropriate fan+heatsink combinations and ordered those.

Leg Stump Simulator

I obtained the stump simulator from Professor Collins’ group and have unlimited access to it until mid-March. At that point, I will work with a PhD student in their lab to establish a week-by-week schedule for borrowing it. To determine the appropriate overall design length for the leg, I made two sizing devices that fit into the bottom of the stump simulator and extend rigidly to the ground. I also identified the dimensions and fastener types used to attach to the bottom side of the iWalk knee brace device.

Challenges

Improved Actuator Design

This redesign was a substantial amount of work. I had been using the 80/20 stock as a crutch, and so now I had to redesign nearly all of the components in some way so that they would fit the new necessary dimensions. One of the biggest difficulties was due to the limited stroke of the ballscrew for the knee. During the early design phase, we didn’t leave enough extra stroke on the ballscrew to accommodate fasteners, attachments, etc.

Time Constraints

The mechanical design and fabrication is taking longer than expected, so I am trying to think of creative ways to reduce the time necessary. I’m also trying to decide how to keep the existing (FVE) version of the leg intact as long as possible in order to let John and Jessica test the control algorithms. 

CNC Toolpaths, Motor Cooling, and Leg Stump Simulator

Individual Progress

CNC Toolpaths

Figure 1: CNC toolpath using 3/8” endmill for ankle ballnut holder

In preparation for the CNC milling operations, I have been following the HSMWorks tutorials on generating toolpaths (Figure 1) for 2.5D milling. I plan to only do the pockets, bushing holders, bearing holders, and hard-to-remember drill patterns using CNC. I will drill the common mounting holes, such as the holes on the corners that have square symmetry, using a manual mill. This will be easy to do because the part can just be rotated/flipped and re-clamped while keeping the table locked (using the mill as a drill press).

Motor Cooling Solutions

Figure 2: Team Integy cooling fan and heat sink for 40mm outer diameter motors

Momentary loads on our actuation system exceed the RMS rated capacity of our motors. However, the RMS load is less than the rated capacity. To help disperse the heat from these spikes in the load, I found combination heat sink-cooling fans (Figure 2) that match the diameter of our motors. These devices are inexpensive and lightweight.

Leg Stump Simulator

I am in contact with Rohan Krishnan in the lab of Professor Collins about the leg stump simulator. Their device uses a commercially available iWALK crutch. This device sits under a user’s shin (leg bent at the knee), and straps to their calf and thigh. In the modified version, a standard pyramid adapter for prosthetic components sits below the user’s knee.

Challenges

Redesign for Simpler Machining

As I plan toolpaths for CNC, I am realizing that some of the parts I designed are unnecessarily complex. For example, the bushings are securely seated in their mounting holes and don’t need to have lips on each end to retain them. This modification means that the parts do not need to be flipped and re-milled in the CNC, saving time.

Also, I am modifying the design of the leg to use the same parts for the ankle and knee actuators as much as possible. I rechecked the calculations of the required diameter for the linear slides, and the ankle actuator linear slide can be downsized while still maintaining a large safety factor.

Leg Stump Simulator Length

Because the pyramid adapter sits below the user’s knee, rather than laterally, the prosthetic-leg side of the user’s body will be taller than the unmodified side. The current solution to this is for the user to wear a platform shoe on the unmodified leg to make up for the height difference.

After we have a system for testing the leg with a non-amputee user, then I may explore shortening the leg to allow use without the platform shoe. This is not as high a priority as getting the weight-bearing parts fabricated.