Takeda, AVI and robotic maintenance

Machine vision and robotics contributions to a regulated pharmaceutical manufacturing site.

April 2025 · RBAI — Takeda Pharmaceuticals Machine VisionMachine LearningCamera HardwareEMA/FDA ValidationPythonConsultingPyTorch
← All articles

Takeda is a biopharmaceutical company with a global manufacturing network. Its site in Lessines, Belgium, is one of the group’s largest manufacturing facilities and produces immunotherapy drugs. Under an RBAI engagement, I contribute to two digital-manufacturing initiatives on that site.

Automated Visual Inspection (AVI)

Takeda lists AVI among its digital-manufacturing initiatives, describing it as a technology that “boost[s] productivity, ensure[s] product quality and improve[s] working conditions” (digital manufacturing at Takeda). My contribution concerns the use of AI-based vision in the next-generation Automated Visual Inspection (AVI) stack on one of the site’s production lines.

Robotic maintenance with Spot

In parallel, I implemented and deployed machine-learning models for object recognition during routine maintenance tasks on a Boston Dynamics Spot robot on site. The work involves setting up the inspection routines, programming directly against the Spot Python SDK, and uploading the processed data upstream.

Takeda in Belgium

RAYA 2025 — Trends in Future Software for Robotics

“Rex as we call him — developed by Boston Dynamics — was awarded for the way we have applied and integrated it into our safety & maintenance routine operations.”

In September 2025, the project received the Robotics Application of the Year Award (RAYA) 2025, category Trends in Future Software for Robotics, presented by ISPE D/A/CH.

Going further

Sources

This article only summarises information already publicly available. Pictures and quoted content come from: