Retrofitting a live robotics network

The situation

A retrofit means changing equipment in a building that's in the middle of running. You get a narrow downtime window, a crew that may be seeing the work for the first time, and station setups that don't always match the drawings. Engineering, safety, operations and supply chain all need to agree on what "done" looks like.

What I ran

Together, these retrofits delivered $62M in value.

• A global retrofit of the wearable safety fleet. This delivered about $50M in savings over three years. At several buildings in Japan the retrofit depended on upgrading the main safety controller inside a two-hour downtime window, so I did those upgrades myself.
• Robotic workcell fixes. Anchoring and joint retrofits worth more than $5M in savings, plus fixes for pinch points, machine guarding and fire risk that avoided a little over $1M in costs.
• A robot firmware update across about 30 buildings with robotic workcells. It added remote interface features and removed the need for vendor service visits, worth about $5M.
• A networking security program that covered hundreds of thousands of mobile robots and crossed several organizations.

On the smaller retrofits I usually managed third-party technician crews directly, often several two-person teams at once. On two of them we closed a safety risk at more than 30 buildings in about a month once parts arrived.

What I changed along the way

A lot of the value came from fixing how retrofits were run, so the next one went better:

• Measuring the value. When I started, nobody tracked what safety retrofits were worth. I worked with safety engineering to define the turnover, labor and liability costs tied to them. That data provided quantitative evidence to keep investing in the team.
• A golden asset. Before a crew moves on to the rest of a building, the first completed unit gets checked and signed off by everyone involved. It becomes the reference for what "good" looks like, which isn't always obvious with rotating teams of technicians and a wide variety of equipment configurations. This became standard practice for the team.
• Technician debriefs before starting. A technical kickoff with every technician before the work begins. Also now standard.
• Integration testing for real buildings. Station setups varied more in the field than on paper, so we added integration testing to catch surprises before the maintenance window.
• Less paperwork. I moved all retrofit documents into one place and made a standard set of templates, covering the schedule, overview, change management, risks and cost tracking.