Navigating the Challenges of Data Capture in Mobile Robotics
In the previous posts of this DIVI series, I’ve explored the defining characteristics of data in the world of mobile robotics and the technical intricacies of data capture. Today, I delve into the often-overlooked yet critical aspect: the common challenges in data capture, and how they can drastically affect the reliability and efficiency of automated systems.
The message I want to bring forward within this article is the fact that sensors made for capturing data are far from optimal. Which obviously has its implications on later steps in the pipeline from Data to making an Impact.
You might assume that nowadays sensors integrated within — for example — cameras are close to perfect if you make the comparison to the cameras integrated into your smartphone. Be aware however that the camera sensors in your smartphone are made for just a few purposes: taking selfies and some nice portret photos. And maybe a few minutes of video clips. They’re certainly not made for capturing lengthy shots of all-weather street scenes or dusty warehouse environments. Even the LiDAR sensor integrated into your iPad Pro is just there for household use cases.
Let’s take a look at a number of issues to consider with these imperfect data capturing sensors:
