25020 – Smart Automation EMMA for Surface Prep

Smart Automation EMMA for Surface Prep

Robotics and Automation

Problem: Surface prep is unpredictable, both due to significant variations in coating thickness and changes in the scope of work or desired level of finish that are only revealed after surface prep starts. This leads to an overly cautious schedule that is expensive and dramatically raises the barrier to entry for automated systems as they need to allow for real-time decision-making and accommodate variability across the workpiece. Also, any solution needs to work using existing infrastructure and must work safely alongside people to allow for parallel workflows.

Description: TAI’s EMMA is a surface prep system featuring vision-based reactive controls and ML. EMMA, unlike traditional robots, uses no pre-programming, no pre-sand scan, and no model of the workpiece being sanded. EMMA continuously monitors her environment and reacts to it in real time. EMMA sees the workpiece, identifies how to reach the surface, and accounts for any movement of the workpiece or herself and continues sanding. EMMA monitors the surface and the environment around her and actively adjusts to avoid colliding with people, the workpiece, obstacles, or areas that don’t need abrasion. A painter uses a tablet controller to direct in real time where and how aggressively EMMA sands, keeping their decision-making in the operation.

Benefits to the DoD: For over 20 years, EMMAs have already achieved:
• Reduced sanding flow by 63%, crew headcount by 25%, and labor time billed to sanding by 73%
• Dramatically improved final paint finish quality
• Eliminated all sanding injuries
• Reduced rework by 80%
• Eliminated schedule volatility
• Increased facility capacity by 250%

Because EMMA does not utilize models or pre-programming and incorporates the painter’s decision making, EMMA:
• works on any work piece
• does any sanding operation immediately
• installs significantly faster than other robotic solutions
• has no downtime when processes or workpieces change
• is usable by new painters with only (1) hour of training

EMMAs are configurable, light enough to deploy from existing infrastructure (scissor lifts/boom lifts), and low cost ($300,000-$400,000 per normal location system), for implementation in whatever quantity best matches a site’s budget/goals for flow and labor savings.

Challenges: The equipment described is the first of (3) phases to achieving fully autonomous sanding. The biggest innovation challenge is improving EMMA’s ML to the point that EMMA correctly decides when an area is finished being abraded and chooses to move to a new area. To achieve this goal of fully autonomous sanding, the data already being collected by EMMAs needs 6-18 months to be processed and trained on to achieve a sufficient performance level to be deployed.

Technical Maturity: EMMA is at a TRL of 9, already commercially available, certified for C1D1 environments, and the equipment shown has sanded (23) production 787s at Boeing with 98% equipment availability.