This Project was probably the keystone and signature project from my time in BR&T. The functional name was the "Mobile Flexible Fixture (MFF)".

The core concept was that we could support a horizontal wing build with a series of discrete mobile platforms. Each MFF had independent drive mechanisms, and closed loop feedback to chase under the attach point being supported. Further complicating the system, each unit was required to have different axes constrained and floating. The root end was tasked with the primary lateral drive, while the tip was used to "steer". 

Functionally, the position was checked at initial load, then the point loads were used as the new zero for each vertical position. Using vertical load cells, the Z axis was adjusted during movement to accommodate uneven factory floors.

The system was demonstrated successfully with a demonstrator box; the learnings and the individual platforms were leveraged in a variety of future projects including wing-to-body joins, predictive shimming, and under-wing drilling.

My individual design ownership was the mechanical design from the X/Y/C axis down to the floor. That meant I was personally responsible for the instrumentation, battery, drive components, wiring and controls cabinet provisioning, as well as the mechanical weldments.

This project represents many of the elements that would highlight the joy and pain inherent in mechanical engineering. The Hexapod Drilling project had the most interesting and most complicated problems I have ever had the pleasure to work on. 

Fundamentally, the problem was was this: We needed a drilling solution for overhead assembly work to remove the risks and costs of manual work, while not requiring massive permanent infrastructure. We developed a mobile base, that could locate itself inside of a fixture using low cost LiDAR, and then localize on part features using trained vision models.

The system replaced traditional XYZ/ABC machine axes with a novel hexapod based solution. Instead we used a W-axis vertical stage, a 6-Actuator Hexapod, and a supplemental C-Axis. From a programming standpoint, this led to a high number of possible singularities, and multiple solution sets for each programmed point and drill vector. Fortunately, we solved these problems before they came to fruition. 

Another key point of novelty that has now become common in industry is the use of a robot tool tender. This allowed high speed automated tool change without the need for a major tool carousel.

I have highlighted this patent because it really exemplifies to me that not all inventions are massive, costly, or involve large teams to improve systems. In this case, we were drilling overhead, and found that drill chip recovery was less effective than initially expected. The drill platform was becoming contaminated with aluminum chips and carbon fiber dust. The clamping nosepiece had a vacuum aperture, but chips were falling before being captured.

The spindle being used was air cooled, and vented to atmosphere after running through the bearing cooling pathways. 

I decided to leverage that waste product to aid in the chip recovery by directing the vented waste back up at the Tool Center Point (TCP). The chips were small, light and dry, which meant that even a minimal stall of their fall resulted in a massive increase in the chip recovery rate. 

We produced the parts using in-house 3D printing, resulting in a marked process improvement with minimal cost or downtime.

This invention is the direct result of a hackathon that my team performed while i was working as part of the Amazon go sensor team. Amazon Go stores worked off of a complicated coordination of vision algorithms, and on-shelf weight sensors. However, traditional stores utilize virtually every surface for displaying and selling merchandise. we set out to add the ability to increase sell-able real estate without adding cost or system complexity. 

Our solution was to create instrumented "impulse strips" that would create a visual indicator for the overhead cameras. By using low power components and simple electronics, we could power our sensors with ambient light via small photovoltaic cells, or a long-lasting battery. The visual indicators were mounted above the sell-able space and outside the visual spectrum, so the consumer experience was not negatively impacted, while still being completely observable by the cameras. To further reduce power consumption we had proposed several additional ideas, like the use of physical flip signage or LCD displays.

Upon joining the Amazon Go Sensor team, one of my first tasks was to reinvent the shelf sensors being used in stores. The then-current shelf sensors were heavy, expensive, complicated, difficult to install, and overbuilt. In addition, they were functionally impossible to adapt for new fixtures or customers. Our task was to improve on each of these aspects without losing the signature reliability that came along with those shortcomings.

Our solution was to create an "Add-On Sensor" that could be applied to Common Off The Shelf (COTS) merchandising shelves. We split the solution into two major components; the sensor hardware, and the structural plates.

The sensors were housed in discrete sealed enclosures, tethered together for reliable wired sensor transmission. We replaced the expensive (but accurate) full bridge load cells with cheap half bridge load cells, using an innovative pairing strategy to output weight and location.

The structural plates were sheet metal, parametrically designed to make them highly manufacturable and easy to create new version for customers requiring custom sizes.

My project responsibility was the structural plates, and later the second version of the sensor enclosures to encapsulate battery power and wireless communication. 

The system had a massive testing and regulatory burden, due to the requirement for food safety approval, use in fridge/freezer environments, and system longevity.

This invention was the result of a strong team effort to drive for constant improvement. In the process of designing the next generation of sensors to use in grocery applications we identified that one of the key requirements was to drastically reduce the power use of our sensing systems. This would reduce heat generation, and make it feasible to go truly wireless. In the process of investigation, we invented a novel load cell methodology based on piezoelectric instead of the standard resistor strain gauges. In our purpose, the key function was to accurately determine weight change, instead of requiring full scale absolute accuracy. We determined we could use calibrated voltage changes to correctly identify and classify weight changes within required margins. And, because the sensor didn't require an activation voltage for measurement, we prolonged system battery life.

However, these new load cells were sensitive to environmental factors in ways that the more industry standard load cells were not. It was my task to create the production assembly package, hermetic sealed enclosure and mechanical load transfer path. 

Composite ply forming has been an area where lots of advancements has happened in the last decade. As more aircraft were being designed with carbon fiber reinforced plastic (CFRP) as integral structural parts, the techniques for laying CFRP needed innovation to keep pace. This invention was one of the incremental steps in that journey, bridging the step between hand-placement and compaction, and automated CFPR tape layers.

The task was to create a way to compact ply-by-ply buildups onto an L-shaped mandrel, with a slight chord-wise arc in addition to the tight L-shape. Before robotic tape laying became a more viable solution, these plies would be placed either by hand, or from a carrier transfer sheet. When pressing these hand placed pieces, it was easy to create wrinkles from dragging a tool similar to a vinyl applicator squeegee. I created a tool that instead created a rolling contact, that was given chord-wise compliance, to allow for even load distribution without generating creases or over compacted edges.

This project was in support of a horizontal wing build prototype initiative. We needed a system that could support the lower panel after delivery from the supplier, then apply a wrap force to help close gaps with the underlying structure before the panel was tacked into place. 

I designed a modular header system, supported by dual scissor lifts, with compliant suction cup headers. The system was designed to support adequately during tacking, drop down, shift to the adjacent rib, and support again during tacking. 

The face of the header was designed to have a linear rail bed that would support either a modular automated drilling solution, or a tool counterbalance. In implementation, either solution would reduce the required overhead work strain.

In order for the panel load, the whole assembly had to drop low enough to clear the wing root, then raise high enough to mate to the wing box. This meant that an extreme vertical movement was required from a very compact lower position.