[ad_1]
Three-dimensional mapping is a technology that is absolutely essential to applications like autonomous vehicles, surgical planning, and virtual reality. However, commercial solutions cost tens of thousands of dollars. This keeps them far out of reach for most hobbyists who would like to experiment with these systems in their personal projects. Moreover, these devices are also proprietary, which makes it nearly impossible to reverse-engineer them and build a lower-cost DIY solution.
Despite these difficulties, mechatronics engineer 9nl set out to build a DIY three-dimensional mapping system. The goal was to rely only on open-source tools so that others could reproduce the work, and keep the cost under a thousand dollars so that it is far more accessible than an off-the-shelf solution. While 9nl was able to achieve this goal, it took a huge amount of effort. Fortunately for us, we can learn from his work and skip straight to the goal of building a cheap mapping tool.
A Raspberry Pi logs sensor data (📷: 9nl)
The system is built around a Velodyne 16-channel LiDAR sensor, which blasts out 300,000 laser pulses per second in every direction. To stay within budget, a used LiDAR sensor was purchased for $400 from eBay. This was paired with a high-precision WHEELTEC N100 IMU module. A Raspberry Pi 5 single-board computer logs data captured by these instruments.
Under the control of a Teensy microcontroller, a motor spins the LiDAR module. This allows it to capture full volumetric coverage of the environment from a single position. The collected data is processed by a SLAM algorithm to produce a three-dimensional map.
Unlike many existing solutions, 9nl’s device does not rely on GPS. As such, it is useful in GPS-denied environments, like caves or parking garages, where other systems would fail.
If this project has sparked your interest, be sure to stay tuned to Hackster News. 9nl promises more videos are on the way, so check back soon to get all the details you need to build your own inexpensive three-dimensional mapper.
[ad_2]
Source link
