Abstract
We describe a method for encoding binary messages in a traffic infrastructure using specially designed rotating targets with barcode-like retroreflective characteristics which encode the data in reflected Lidar intensity patterns. These Lidar-readable Rotational Barcodes can be installed as elements of a traffic infrastructure for Infrastructure to Vehicle (ITV) signaling, and they can also be installed on vehicles to provide identification (e.g. for an emergency vehicle) or function for Vehicle to Vehicle (VTV) signaling. Since many self-driving vehicles already use Lidar for mapping the surrounding location and avoiding other vehicles and obstacles, our approach does not require any additional equipment on the vehicles themselves. The paper begins with a description of the the physical construction and operation of a Lidar-readable Rotational Barcode. We then develop a two-stage detection algorithm based on windowed standard deviation of the reflected Lidar intensity followed by comparison with neighboring bearings. Finally, a decoding algorithm is described that first implements median filtering and compensates for an “edge effect” before making an optimal assignment of bit numbers to bar widths. We give details of a proof-of-concept experimental setup and outline the results of our detection and decoding algorithms on data gathered by a Lidar-equipped wheeled robot in a lab environment.
