Better Eyes; Better Robot Outcomes

With better eyes, we get better robot outcomes which lead to more trust in our automation systems

Customer visits provide an excellent learning possibility. Recently I met a company that makes robotic automation systems for a variety of industrial automation applications, including palletization and depalletization for the food and beverage industry. The primary challenge facing them was to automate the task of beverage palletization. I noted down some of their difficulties;

• Identify cases, with and without labels or specific markings
• Avoid tricks that cardboard bulges, overzealous flaps, and damaged corners play on less sophisticated robot vision systems
• Reliably describe what to look for when carton sizes regularly change to marketing needs
• Assure that beverage cases, once shrink wrapped, don't have "tails" that get snagged in machinery

The common denominator? The smart people in the R&D department at the customer realized all their automation problems were related to machine vision systems.

Everything is a vision situation. With better eyes, we get better robot outcomes which lead to more trust in our automation systems.

A well-designed vision system can alleviate many, if not all, of the problems outlined by the customer. An ideal machine vision solution allows;

• SKUs to be verified by sight independent of markings and labels
• Mitigates against object orientation issues
• Deals with creative marketing departments that need to modify packaging

However, the customer had not yet found a solution that checked off everything on their list of requirements:

• Human-like 3D imaging
• High acquisition and data processing speed
• Software APIs and tools
• Ease of use
• Robustness for use on factory floors