Microscopic Ore Rover

1.0 Summary

The Wireless Microscopic Ore Rover, or WMOR, is a remotely driven robot built to investigate the regolith of a new planet and identify ore worth mining. It carries three things that matter: an onboard microscope with a sample collection system, an FPV camera for driving it by sight, and an ESP32 vision system that can track a target and trigger collection on its own.

It was a team project. My contributions were the camera holder, designed and modeled in Fusion 360, and the microscope's magnification testing.

2.0 Objective

The brief was to build a rover that could explore a landed planet, sample its surface material, and characterize what it found, as a stand-in for the kind of extraterrestrial resource survey that has to run without a person on site. That set the three subsystems: get there and see, collect a sample, and look at it closely enough to tell one material from another.

3.0 The Rover

We started from a base rover and added the parts that made it useful: the microscope and sample collection assembly, the FPV camera for teleoperation, and an ESP32 camera for automated target tracking. Each one had to earn its place on a chassis that still had to drive.

4.0 Microscope & Sample Collection

The microscope stacks a lens and a CCD shell, a sample holder set at a fixed working distance, a thin film to keep particles out of the lens, and a particle concentrator on top to gather them into one spot to view. An LED at the inlet lights the sample, driven from an Arduino through a potentiometer so the brightness could be tuned.

Assembly order mattered. The film goes on before the concentrator so particles never reach the lens, and the whole stack was checked against the video receiver before it went onto the rover in place of the arm.

5.0 FPV Teleoperation

The FPV camera is what makes the rover drivable when you cannot see it directly. We tested two lenses. The standard 120 degree lens gives a clean front view and easy control but little sense of what is beside the rover. The 180 degree fisheye shows far more of the surroundings but distorts heavily and makes distances hard to read.

We kept the 120 degree lens. For teleoperation the natural, low-distortion view was worth more than the wider field, and we confirmed it worked in the dark under the rover's own LED.

6.0 Automatic Target Tracking

The ESP32 camera streams video over WiFi and runs color tracking on it. It locks onto a red pole, estimates the distance to it, and once the rover is close enough it triggers the ore collection sequence. That closes the loop: the rover can find a marked sample and collect it without a driver steering the last stretch.

7.0 Camera Holder Design

The camera holder was mine. It had to fix the camera at the right height and angle, clear the microscope stack, and mount to the existing chassis without new holes. I worked it as a Pugh chart of options first, then modeled the chosen design in Fusion 360, checking it in orthographic views and detailing the cutout in a sketch before it was made.

8.0 Testing

I ran the magnification testing on the finished microscope, both the on-screen digital magnification and the optical magnification from a section analysis, to confirm it resolved detail well enough to actually classify a material.

The check that it worked was telling two particle types apart. Under the microscope, sand shows rough edges and smaller crystals of uneven size, while salt shows larger, more structured crystals of uniform shape. Being able to see that difference on the video feed is the whole point of the instrument.

9.0 Lessons

Most of the hard moments were not design, they were signals. The FPV feed came in full of static until we found we were on the wrong frequency band and channel; once matched it was clean. A set of motors failed and had to be taken apart and resoldered. First-year or not, the lesson that stuck was that an integration bug hides as a hardware bug until you check the settings.