The Fast Traversing Autonomous Rover for Mars Sample Collection is a NASA funded project at WVU. The purpose of this platform is to test Mars rover autonomy path planning algorithms and acting as a test bed for scientific payloads.
Although this rover has changed mechanical leads 4 times in the 4.5 years this project has been ongoing. I have been working on this project for 3.5 years and the Mechanical Lead for the past 2 years. I have seen this assembly go from a very crude and unrefined model to an assembled and driving platform.
There is still programming and further integration to do on this 360 lb beast, but it's nearly complete and I greatly looking forward to all systems go!
When I joined the project, the CAD of the rover was in a geometric state with many manufacturing and assembly details not included yet. Over the course of a year and a half many details were added by 4 other engineering students and myself.
We focused on refining the design for manufacturing components essential to the robot's functionality first. The Central Frame was the first component to take shape in the construction of rover. This interlocking set of plates provided an easy to manufacture core structure with ample mounting points for future modules. One feature that made this design possible was the use of the outer plates to tie together the central frame.
The next major assembly to gain definition was the Drive System. This design focused on maintenance by allowing a user to remove the motor and its adapting hub without removing the wheel and bearings from the wheel flange.
The Steering, Control Arms, & Linear Servo Mounting assemblies were all developed in parallel. In doing so we managed to aqcuire a large range of motion while avoiding collisions and components complementing each other well.
The Steering Assembly is by far the most complicated in the rover. The space available for it was small so that the rover could drive through a standard elevator door while maximizing the suspension travel. Incorporating the spring suspension, off-axis driving motor, motor controllers, rotary encoder, & depth sensing into such a small space proved to an interesting endeavor.
This section of the robot was the most subject to change throughout the process. Despite five drastic iterations in the design phase, there were significant modifications post-manufacturing to better accommodate electrical devices, many of which changed after testing.
The Control Arms were primarily designed so that the 400 lb linear servo would be able to lift the rover while maximizing the wheel travel. Being able to drive through an elevator door and having limited internal space for the servo greatly contrained the system.
The length of the control arms were determined based on wheel placement with respect to the rover frame. The outer distance from wheel to wheel ended up being 44-46" to remain easy to load into the elevator. While the size of the internal cavity restricted us to a 2-4" stroke linear servo with some breathing room. These two length dimensions determined the 400lb linear servo requirement with a healthy factor of safety. These constraints were used as some of the inputs for a MATLAB program plotting out the wheel travel in the resulting four bar mechanism.
Finite Element Analyses were conducted on the control arms to validate the strength of the design due to stress concentration concerns. Several iteration were made, and it quickly became apparant that the secondary brace in the load bearing control arm was needed to avoid a catastrophic bending moment at the base of the arm.
The Linear Servo Mount was an rather cumbersome design to produce. Many of these other assemblies either had a lot of freedom to define their interface with other parts, or had a good foundation to work from. This assembly had niether of those.
The central frame does not have symmetrical fastening points, and the actual mounting of this assembly was an afterthought during the central frame design. Which this central frame had already been manufactured at this point. So, a rather creative solution fruited as a result.
A MATLAB program was written to compute the travel of the wheel as a fourbar mechanism. The elevation and recession of the base for the linear servo where altered to maximize the travel of the control arms, whose length also varied. Only after running many scenarios in this program, were the servo location, servo stroke, and control arm length finalized.
The monting took shape around this general location to place the servo base. This resulted in a flat bar spanning the corner, and strange hole placement and adapting plates to tie in this bar into enough points on the rover frame.
With these assemblies designed and initial fitment fix. The first "Leg" Prototype was manufactured and assembled.
Although this worked, there were some improvements to be made to simplify machining & to reduce unintentional slop due to hardware selection.
All four leg assemblies were manufactured after minor revisions to the parts and selecting different hardware to compliment.
A metal stand was constructed to let the rover frame rest on while legs were being installed. This has also proved useful for maintenance on the leg assemblies themselves.
This past summer the rover was taken for a test drive to confirm functionality. Although many of the sensors at not added in this test drive.
There were some concerns among team members, but many of the witnessed issues were due to either fasteners that weren't tightened or using a skid steer drive controller for an independant steering system.