The Challenge
A brake pad manufacturer in the Cleveland area invested in a Mitsubishi PLC, HMI, a UR collaborative robot, and a Cognex vision system to automate feeding brake pads into a machine.
Despite having quality parts in place, the lack of proper integration meant the automation struggled with reliability and performance. This was not a long-term solution.
- The vision system frequently misidentified parts or failed to switch jobs.
- Operators had to step in constantly to reset cycles or press hidden buttons.
- The robot sometimes miscalculated its pick points, missing parts and crashing.
- Stacks up to six layers high created complexity the system could not reliably handle.
Instead of increasing efficiency, the system drained operator time, reduced throughput, and left the customer questioning their investment.
Partnering with Axis
The manufacturer asked Axis to analyze the system and determine how to improve the system.
What Axis found was a familiar issue in complex automation projects: individually capable components that weren’t integrated properly. Parts alone don’t make a reliable automation system. What’s really required:
- Clearly defined solution. What does good look like? What is the goal?
- What are the problems?
- What resources are available?
- An automation partner that will take the time to understand the problem, develop a solution and the ability to implement it.
The Axis Approach
Key Improvements:
- Vision System:
- Re-mounted and refocused the Cognex camera system,
- Optimized lighting, added a blue filter
- Consolidated 80+ job files into a single streamlined job file for easier recipe management.
- Calibration:
- Built a robust calibration board
- Created robot-assisted calibration routines
- Integrated calibration requests into the HMI
- Provided clear operator instructions.
- PLC Communication:
- Redesigned the “soft bit” communication sequence to load, request, and verify camera data reliably, reducing cycle time and eliminating errors.
- Robot Program:
- Consolidated multiple UR robot programs into one, position corrected X/Y coordinate handling (including negative values), integrated angle data, and added force-sensing to adapt to varying part heights.
- Operator Interface:
- Redesigned Mitsubishi HMI screens to display cycle status, recipe parameters, and error prompts in plain terms, improving operator clarity and confidence.
- Additional Sensing:
- Added a Keyence laser sensor to measure stack height in real time, ensuring the robot always picked from the correct layer.
This work transformed the cell from a fragile system into a robust, self-correcting system.
The Proof in Testing
At project completion, the customer tested the system aggressively. They moved parts mid-cycle, removed pads from stacks, and rotated pieces in odd orientations — scenarios that previously would have caused downtime.
This time, the robot adapted seamlessly. It identified the parts correctly, adjusted pick positions, and completed cycles without a single crash.