LOST CABIN
A proving ground for autonomous exploration.
Somewhere in Wyoming's Big Horn Mountains lies the story of a historic gold discovery whose location was lost more than a century ago. Baryon will use the search as a real-world test of a larger idea: autonomous machines that can explore, measure, reason about, and understand the physical world.
A 160-year-old search problem.
In the 1860s, a small party of prospectors reportedly struck rich placer gold in the Big Horns, built a cabin, and were driven out — leaving a discovery no one has verifiably relocated since. What remains is exactly the kind of problem modern exploration still struggles with: an enormous search area, difficult terrain, incomplete historical information, sparse observations, and multiple competing hypotheses.
Traditional exploration requires people to repeatedly traverse difficult terrain, collect measurements, return with the data, interpret it, and decide where to investigate next. Baryon is asking a different question: what happens when the exploration system itself can make that loop autonomous?
One mission. Multiple machines.
Lost Cabin is not a drone project or a robot-dog project. It is a coordinated system: fast aerial machines that survey large areas, and ground machines that investigate what the air finds. Each platform is ordinary on its own. The value is in the loop between them.
Air — Baryon Aerial Scout
A VTOL fixed-wing survey aircraft that flies systematic autonomous grids and builds a georeferenced picture of terrain, geological structures, lineaments, surface alteration, vegetation anomalies, and magnetic anomalies — the map that decides where the ground team goes.
- RGB mapping camera
- Multispectral
- LiDAR
- RTK GNSS
- Precision IMU
- Terrain following
- Magnetic survey payload
- Towed magnetometer
Ground — Baryon Ground Scout
A quadruped geophysics platform that walks to the anomalies the aircraft flags and measures them properly — denser data, closer to the rock. It is the machine that closes the gap between remote sensing and drilling.
- LiDAR
- Stereo depth cameras
- RTK GNSS
- Magnetometer
- Gamma-ray spectrometer
- EM conductivity
- XRF
- Multispectral
- Environmental sensors
- IP/resistivity deployment — in development
- GPR — in development
- Autonomous sampling — in development
Autonomous exploration loop.
Define the space
Historical maps, geology, elevation data, satellite imagery, and known occurrences define the initial search space.
Fly the grids
Autonomous UAVs perform broad-area LiDAR, imagery, multispectral, and magnetic surveys.
Find the anomalies
Onboard and cloud models identify geological structures and anomalous sensor responses.
Assign the ground team
Ground robots are automatically assigned to the highest-value targets.
Measure up close
Quadrupeds collect denser geophysical and geochemical measurements at each target.
Update the model
The combined observations update a continuously evolving geological prospectivity model.
The next mission is generated from what the previous mission learned.
The physical world does not speak in one sensor.
No instrument detects gold directly. Gold exploration is fundamentally an inference problem — many weak physical signals, combined carefully, become progressively stronger geological evidence. That is why every Baryon platform carries more than one way of asking the ground a question.
MAGNETICS
Maps structures, lithological boundaries, intrusions, and alteration.
LiDAR
Produces high-resolution terrain geometry and identifies subtle surface features.
MULTISPECTRAL
Characterizes surface materials and alteration signatures from their spectra.
RADIOMETRICS
Measures natural gamma signatures including potassium, uranium, and thorium.
EM
Maps electrical conductivity and subsurface structures.
IP / RESISTIVITY
Higher-resolution ground investigation of potential mineralized zones.
GEOCHEMISTRY
Ground robots can ultimately collect and analyze rock and soil observations.
The product is not the robot.
It is the system that coordinates them.
Baryon is not building a portfolio of disconnected machines. The long-term technology stack is an orchestration layer that lets autonomous machines work as one exploration organism:
- perceive their surroundings
- understand mission objectives
- allocate tasks across the fleet
- coordinate with other machines
- react to new observations
- collect structured physical data
- update models of the environment
- decide what to investigate next
- escalate high-value targets
- return, recharge, redeploy — automatically
Why start here?
Real terrain
Mountainous, irregular, and physically difficult — no laboratory floor.
Real uncertainty
The target is not precisely known. The system must reason, not follow waypoints.
Real science
Success requires geology, geophysics, mapping, and inference working together.
Real autonomy
The machines must navigate and collect useful data far from their operators.
Real feedback
Every mission creates information that can change the next mission.
Real economic analogue
The same architecture applies directly to modern mineral exploration.
Lost Cabin is not the destination. It is the benchmark.
From exploration to autonomy at industrial scale.
Lost Cabin is a deliberately constrained first problem. If machines can autonomously explore a mountain range, collect scientific observations, reason over uncertainty, coordinate follow-up missions, and improve their understanding of the environment — the same fundamental architecture extends far beyond exploration.
Lost Cabin
Autonomous geological exploration
Mineral Exploration
Autonomous reconnaissance and target generation
Mining
Autonomous drilling, sampling, surveying, and material movement
Infrastructure + Industry
Autonomous physical work across remote and industrial environments
The Physical Economy
Networks of intelligent machines performing economically useful physical work
Baryon is building toward a future where intelligence is not confined to screens. It operates machines, moves through landscapes, interacts with matter, and performs work in the physical world.
Mission telemetry.
Metrics populate as the field program executes. Nothing here is fabricated — until a machine measures it, it reads as pending.
Program roadmap.
THE MOUNTAINS ARE THE LAB.
Lost Cabin will be Baryon's first field test of machines that can autonomously explore and understand the physical world.