This DIY Sensor System Maps Caves in 3D

I’ve done a fair amount of caving and I can tell you from experience that being deep underground is about as isolated as it is possible to be. You could be stranded in the middle of the Pacific Ocean and at least GPS would work, but no signals penetrate through the Earth when you’re in cave. That makes most technological navigation systems useless, which is why 9nl built this DIY sensor system to map caves in 3D.

There are commercial systems built for this task, but they cost tens of thousands of dollars and only a tiny percentage of caves have been mapped with them. Most of the time, cavers have to rely on rough hand drawn 2D maps, which aren’t very useful when “up” and “down” are critical directions in a cave. Heck, hand drawn maps usually aren’t even an option, which forces cavers to navigate by memory or markers left by others.

9nl’s system is basically a handheld lidar unit for SLAM (Simultaneous Localization and Mapping), but without any external assistance. It spins a 2D lidar sensor assembly (a Velodyne VLP 16), which itself has an internal spinning sensor unit. So, it spins in two directions, covering all angles, constantly looking at the distance to the surrounding walls. A Raspberry Pi 5 running ROS2 takes that raw sensor data and creates a 3D map. It is similar to 3D scanning, but inverted.

That would be enough to scan all the walls in view if the system were stationary, but it wouldn’t do much good for continuous mapping of an entire cave. For that to work, the system needs to know where it is and how it is moving through space. 9nl achieved that with an IMU (Inertial Measurement Unit) and he’s quick to point out that the IMU quality is crucial for getting good results.

All of those components went into a portable enclosure with a battery, so 9nl can carry it through a cave. Initial above-ground testing seems promising, but the real gauntlet will be an actual wet, muddy cave full of irregular surfaces that 9nl will have to scramble through.