Case Study: Acquisition, Technology & Logistics Agency (ATLA) achieves colorized far‑infrared imaging and autonomous AGV navigation with MathWorks MATLAB

Acquisition, Technology & Logistics Agency Applies GAN to Far-Infrared Images and Generates Color Images

Japan’s Acquisition, Technology & Logistics Agency (ATLA) researches far-infrared imaging to enable stable environment recognition, but faced low visibility and sensitivity to temperature changes that made object identification difficult. To address this, ATLA used MathWorks tools—primarily MATLAB and the Deep Learning Toolbox—to develop and test new deep learning workflows.

Using MathWorks’ MATLAB environment, ATLA implemented an image-to-image translation (pix2pix) GAN, calibrated far‑infrared and color cameras to create training data, and trained models to generate highly accurate color images from infrared inputs. The MathWorks-based solution improved visibility and object detection performance and was deployed on an automated guided vehicle (AGV), enabling autonomous movement in low-light conditions.

ATLA