Project Examples
Manufacturing Renovation Project Management
Project Portfolio Management:
work with higher management to discuss and select a best-practice project management plan and specify procedures to track and report project progress
work with project managers to draft template documents for project progress tracking
guide and train managers in performing cost-effectiveness analyses, setting up reporting templates, and technical writing
Project Initiation
work with production managers, engineers, and line leaders to analyze/optimize the operation processes and identify automation opportunities
provide technical advice/consultation to the end-user team, research relevant technologies and available resources, and perform simulation or proof-of-concept research to manage risks when necessary
connect with and invite exterior system integration vendors to seek automation renovation project proposals, clarify requirements, and advise/communicate with them regarding risks versus profitabilities
Project Management Implementation
evaluate all submitted project proposals and internally discuss to select the most promising one to pursue; communicate with the vendor to iteratively revise the project plan to develop a binding contract
technically ensure communication between the design team and the end-user team, track the project progress, review and prove the system design on behalf of the end-user team
perform acceptance procedures, review documentation, and help train the end-user team
Manufacturing Research & Development Project Technical Services
Risk Management, Proof-of-Concept Research
investigate and explore renovation options with the operations team, in order to evaluate available automation technologies and the risks of adopting them
work with the project manager to research and simplify the situation, aiming to pinpoint the risk factors
collect data for the simulated situation to test the performance of the proposed algorithms and to determine system components/specifications, based on an optimized cost-effectiveness criterium
Example: Cobot, Computer Vision, Force-Feedback Control Tech Status Research Demo
The characteristics of cobots (or collaborative robots) that allow them to work with human operators in a shared workspace make them attractive for existing production line renovation.
Compared with a human operator, a robot can work tirelessly with higher repeatability and consistency, and its weak point (less intelligence) can be mitigated with advancements in sensing technology.
This video to the right demonstrates how well an articulate-type cobot with the currently most advanced 3D Computer Vision sensing technologies, combined with a Force-Torque Sensing Feedback intelligence — both of a reasonable cost — can perform traditionally challenging pin-hole high-precision assembly tasks.
Example: Computer Vision for Robotic Power Coating Cell Operation
A powder coating line built in the 1990s had a 2-robot painting cell that could paint ~90% of the parts (of about 20 models) going through the conveyer. It used the light curtain profiling technology to recognize the registered parts of the 20 models and call their corresponding robot painting programs.
The project uses computer vision to replace the Light-Curtain ID Stand without having to change the rest of the production line. The benefits of the project include: (1) enhanced productivity to enable all parts to be painted by the robots; (2) enhanced control of painting quality; and (3) reduced waste from scrapped parts.
Robotics Research & Development Technical Services
Machine/Equipment R&D
industry market research to find out the ideal product specifications and the trade-offs among the key performance factors
key components research to tabulate available resources (specs versus prices & lead time); analyze the data to select the key components, pilot-research in proof-of-concept if necessary
system overall mechanical design, arrange the key components to estimate the required spaces for the camera sensor and optical subsystems
camera sensing subsystem design: sensor chip selection, field-of-view, focus range, aperture, field-of-depth, lens selection, protection dorm, focus-adjustment mechanism, LED lighting
mechanical design: experiment/estimate normal friction with the given seal mechanism, detail out the motor power gear transmission/reduction mechanism, detail out seal mechanism with the profile constraints, and detail out the body/connection parts to ensure PCB circuits room
generate parts drawings (detailing out GD&T) for fabrication quotations, fabrication support, verify received parts for quality check, guide assembly process, and test system performance
generate assembly and maintenance documentation while working with end-users to collect feedback, R&D phase ends & maintenance/improvement phases starts
Sensor Data Processing using Artificial Intelligence (Machine Learning) Technologies
work with chemistry scientists to determine the sensor array configuration, and work with electrical engineers to detail out a test and data-collection scheme
design basic sensor data process scheme: removal of outlines, noise filter based on frequency analysis, buffer size selection to separate gas composition variation from sensors' baseline drift, design algorithms to estimate baseline drift profile
perform initial sensor array data exploration in the hyperdimensional space, select visual presentations (such as spider charts) to communicate with chemistry material scientists, help sensor invention research and refine sensor array configuration
with the given sensor array configuration, design experiments of a series combination of target gases' concentration to collect data, try different machine learning algorithms to select the most promising one (the most effective machine learning algorithm was Neural Network) to further pursue
work with chemistry material scientists to design and implement experiment series, and machine learning process the training data offline with the defined sensor array configuration and settings
transfer the trained gas detection algorithm to an embedded computer system for real-time testing