Friendly Robots Company
Mechanical Engineering Internship
Timeline: February 2020 - August 2020
Skills used: CAD (Fusion 360), ROS, soldering, sheet metal manufacturing
Friendly Robots Company is a robotics startup that specializes in autonomous vacuum robots. As a mechanical engineering intern, I mainly worked on designing and prototyping the robot's end effector where the rotating bristles and vacuum head are located with Fusion 360. On the side, I worked on some electronics for controlling the circular vacuum brushes that involved coding in the ROS environment. The result was a fully sheet-metal based end effector using a four-bar linkage mechanism that was in-house manufacturable, cheaper, and more rigid than their previous milled design.
Situation: It was too expensive to build more robots
The robotics startup was focused on building more robots to ship to their customers, yet the mechanical configuration of their prototype took too long of a turnaround time to fabricate such that it made large-scale manufacturing unsustainable given the available resources. One way they suggested solving this problem was to make their robot vacuum end effector, which currently required sawing off a metal bar from a stock wheelchair chassis and welding on a CNC-milled vacuum mount, out of only sheet metal that can be printed and folded in bulk. Alongside another intern, we were tasked in designing such an end effector piece.
Existing end effector mount that required sourcing of different parts from different sources fabricated using different manufacturing techniques (3D printing, milling, sawing). We wanted a solution that simplified part sourcing and manufacturing.
Task: Iterating on an unconventional design
The design process involved 5 main steps: research, brainstorming, low-fidelity prototyping, high-fidelity prototyping, and assembly documentation. In between each step was some form of testing and iteration that will be explained below
Research
This involved gathering information on what constraints I had to account for and problems with the current design I had to solve for. Some of these issues include limitations in how much the vacuum could retract, dependence on a manual strap to retract mechanism, and that vacuum brushes kept falling off. In addition to the sheet-metal material , keeping the footprint of the end effector to a minimum was another major constraint for this project. I also gathered the dimensions of the chassis for use in the prototyping phase.
Sketches of the dimensions of the vacuum to be mounted
Brainstorming
The goal of this part of the design process was to come up with as many ideas as possible no matter how crazy they were and gather feedback on each one so that we could move on to low-fidelity prototyping. The other intern and I did not have any experience with sheet metal design so it was useful to discuss amongst ourselves and with other engineers within the startup what seemed plausible and what did not – it was all trial and error at first. After considering various ideas, we eventually decided on a four-bar linkage system that theoretically allowed the vacuum to move vertically while maintaining an orientation parallel to the ground.
Sketches from our first brainstorming sessions
Low-fidelity prototyping
We created 1:1 scale cardboard prototypes of our sketches to test out the feasibility of the four-bar linkage mechanism we converged on, specifically if it allowed the vacuum to move parallel with the ground and if it were possible to fabricate all parts with just 2D sheets of metal for high-fidelity prototyping. We learned that the mechanism did move as expected when unloaded and was mostly possible to create with just flat sheets of cardboard; however it was not strong enough when loaded with the vacuum and the connection to the vacuum head was still not fully realizable with completely flat parts.
Cardboard prototypes based on our initial ideas
We decided to create low-fidelity prototypes with plywood connected with hot glue and some available nuts and bolt fasteners to see if the increased rigidity of this material will give us a better picture for how the mechanism can hold up the vacuum when manufactured with sheet metal. This involved CAD on Fusion 360 with all of the dimensions gathered during research. This plywood prototype enabled us to test dimensions on a cheaper manufacturing method before committing to a sheet metal design. During this process, we also made structural improvements to the design due to feedback from engineers within the company as well as third-party manufacturers working with us to increase strength such as adding a horizontal bar connecting the two four-bar linkages lifting each side of the vacuum as well as gussets at angled connections.
Plywood prototypes building on what we learnt from cardboard prototypes
High-fidelity prototyping
Once we were satisfied with the fit of these prototypes, we obtained 14 GA steel sheet-metal cuts of our design from a third-party manufacturer and obtained our own hydraulic press to fold these in-house. We discovered a lot of flaws in our sheet metal designs after trying to fold them ourselves, such as the fact that box geometries cannot be folded to form a closed volume using our hydraulic press. We had to iterate a couple more times to make sure all components can be folded correctly.
We assembled the entire design with nut & bolt fasteners for mobile joints and rivets for fixed joints. The result was like we had envisioned with the CAD and was strong enough to hold up the vacuum head and bristles, though there was still a lot of friction with the joints and it was overall very heavy.
How our design changed knowing how box geometries could not be folded with our hydraulic press
CAD model of first sheet metal prototype
Fully assembled second iteration of the sheet metal prototype
Assembly documentation
I created a visual assembly manual so that other engineers in the company can build more of these prototypes when I leave at the conclusion of my internship. This involved creating instructions for the folding of individual sheet metal parts as well as how they all were combined with the appropriate fasteners. The manual involved minimal text and was well received by my co-workers trying to assemble their own robots.
The assembly manual uses screenshots from Fusion 360 highlighting what parts are modified in between assembly steps
3 Key Contributions to the Team
1. During brainstorming, I came up with the four-bar linkage idea inspired by a kitchen shelf cabinet.
2. Managed large CAD project dependencies using a hierarchy of subassemblies so that my work with my fellow intern did not conflict.
3. Optimized sheet metal geometries using design for manufacturing with regards to sheet metal, which involved separating parts when necessary to decrease the amount of folding required and maximizing surface area of contact.
Outcome: Sheet metal vacuum mounts are possible
We ended up with a solid, sturdy end effector mount design fully made of sheet metal that could translate up and down as required. The biggest accomplishment was showing that it was indeed possible to design a vacuum end effector mount fully with bent sheet metal parts that cost only around $200 total per robot. However, our result was not perfect because there was too much friction between joints and was too heavy to lift manually. A possible solution to this would be to make use of a transmission system (perhaps a pulley system) to minimize the force required to lift the end effector and allow the lifting mechanism to be motorized, however I have not been able to implement this after the internship ended.
CAD model of second iteration of sheet metal prototype