The Stone Mine Unmanned Ground Vehicle is an Alpha Foundation funded project at WVU. The purpose of this project is to construct an autonomous robotic surveyor for underground mines.
A Unmanned Aerial Vehicle (UAV) system is ideal for scanning the 40 ft tall and 100 ft wide underground pillars. However, there is a lot of ground to cover and a lot of operation time to document it all. That is where the support Unmanned Ground Vehicle (UGV) comes into play. This UGV provides the drone with power through a tether cable to extend flight time. This tightknit relationship between the robots led to the nickname of Rhino (UGV) and OxPecker (UAV), after the symbiotic relationship between those two animals in the wild.
The document below summarizes the design, manufacturing, and integration process that has occurred over the 2.5 years of working with this 400+ lb beast!
Rhino and OxPecker make their debut at the WVU Innovation Hub Dedication Event.
Over the course of a summer, I lead two other undergrad students in the design of this UGV. We went through a thorough design process that aims to remove bias and better narrow down the design criteria in an efficient manner.
This process began by minimally defining our problem and goal. While being sure to not include solutions in the definitions made. In this application, we realize that long-term mine safety inspections are a time consuming, highly repetitive, and dangerous job for humans due to deteriorating mine structures. Therefore, we are going to create an means of autonomously surveying mine support columns with LIDAR for cracks, degradation, & surface discontinuities to monitor structural integrity.
This is a sample LIDAR scan from a pillar in the Test Mine.
And this is the resulting feature recognition map generated by a program made in the WVU Mining Department.
The next step revolved around our minimal system requirements and what features are needed in order to achieve those system requirements. It was clear that a drone equipped with LIDAR was needed to scan these 40 ft tall columns. However, operation times reaching close to 8 hours are needed to make the most effective use of section closures. This extended flight time requirement gave birth to the support rover, which provides power and a means of transport through the mine.
These few requirements spawned a number of features:
Drone tethered to Rover for power
Method of securing Drone to the landing platform for transport
Rover must fit a standard freight elevator door (48” wide)
Rover capable of driving over small debris (Simple passive suspension & 6” minimum ground clearance)
Rover robust enough to withstand small falling debris
Simple assembly for ease of manufacturing & maintenance
Able to operate in dusty and moist environments
With these features in mind, a handful of core components were selected. Such as drive motors, batteries, a suspension system, sensors, computer, power management, lights, and typical rover equipment to make the system usable. This is where we started making a geometric model of the rover to coordinate parts and solidify the general structure. We treated this robot basically as one big adapter between all the hardware components that we had selected. Eventually we settled on a wheeled skid steer rover with a split chassis body with limited articulation for the suspension.
Once the form began to take shape, more parts and details to the structure were added. In parallel to designing, we planned for:
Assembly (including the tools & fasteners used to assemble)
Cable/Plug Management (ample room for connectors & wire routing)
Maintenance/Modularity (easy to work on & remove components as needed)
Manufacturing (minimizing cost & complexity)
Sensor Obstruction (maximizing the use of visual sensors & comms equipment)
Tolerance & Work Space (making the design tolerant to the work environment)
The design we came to utilized a primarily waterjet chassis. Of the approximately 160 mechanical parts, 117 are water jet at the WVU Innovation Hub (our in house manufacturing). The central frame is welded together, using the 5 CNC parts and a few components as references. While the outer skin is pop riveted on this frame.
As the design continued to take shape new features were added:
Transportation locks
Asymmetric Battery Layout
Refined Lighting Layout
External Monitor Mount
Improved Cable Management
Improved View Angle Coordination
One of the undergrad students conducted a few FEA simulations on the chassis before finally we sent out the design for manufacturing at the end of the summer. We simplified the components involved based on where we saw the least material for the given load. Some of these included designed failure points. These scenarios included: 2 m/s Front Collision w/ Bumper, a 1 m. Drop (for Rough Terrain), Carry by Bumpers w/ Locks, & Roll Over (Roll Cage Only).
The water jet plates came out beautifully, and were welded together with a few components and machined parts as references to ensure fitment. All of the robot frame is made of 6061 Aluminum. While the wheel hubs are made of 301 stainless steel for strength.
The frame, roll cage, and bumpers are all painted in a Black Gloss from KBS Coatings. While the skin is painted in a "CAT Yellow".
This excellent paint did have its own learning curve though. As myself and another grad student applied the paint, we quickly figured out that we needed to thin it out more than we had and to etch the aluminum for a good bond. This resulted in the skin plates being remade, but the rest of the rover parts were cleaned, etched, & repainted.
Parts for the rover lined up beautifully, and the progress was well. However, a minor set back occurred when I managed to break the pop rivet tool since the size of steel rivet used was too much for it. The silicon caulking was a fine addition of environmental protection and really helped the system come together.
We discovered a few issues after the rover had been assembled and driving. Thus began the woes of integration hell.
First of all, these motors are much larger than what we have worked with before. The back EMF of the system is just as massive and would force our motor controllers to shutdown. This was eventually allieviated with an in-rush current limiter.
Another issue was the size of the computer used, we had ultimately used a different computer than intended. However, this was now too large of a motherboard to fit the other components. I sought out to find an ITX motherboard to replace it, only for it not to post when put together. I discovered that somehow a pin in the CPU socket had been bent down into the socket! Thankfully a local repair shop was able to get the pin back in place to save our computer!
Another issue was more so a short cut in manufacturing that came back to bite us. We had opted not to buy the expensive brooch set that matched our motor shaft and key size. Instead, we utilized multiple set screws into the keyway. This only worked for so long, and evnetually lead to the strenuous extraction of these wheel hubs. A new section of shaft was welded onto the assembly after machining down the existing hubs. This ultimately gave us the perfect geometry for reliable performance and easy removal.
An additional problem in this grad adventure came as a result of a few circumstances. A motor that we realized was too fast for our application and therefore not enough torque was paired with very aggressive closed-loop control for wheel position. This led to the untimely demise of two motors, and hastened the replacement of these motors with their better geared cousin. (increasing torque by 30%~)
Despite all of these integration pains, we managed to get this robot together for the WVU Innovation Hub Dedication Event! The new computer, improved wheel hubs, in-rush current limiters, and back-up motors enabled us to drive the system down to its show case with ease!
The presentation poster I threw together turned out well. It helped a lot when explaining the project to investors and enthusiasts alike.
Rhino is currently going through the final integration of sensors and test drives across campus to check our fixes. There are still some minor tweaks to be made to the base system, but we have come very far on this UGV from the 2+ years since this project started.