« Back To Robotics Editorials
GrayMatter Robotics Logo

Member Since 2020


At GrayMatter Robotics, we tackle manufacturing challenges with advanced robotics and proprietary GMR-AI™. Our automated surface finishing solutions optimize high-mix applications, boosting efficiency, reducing setup times, and cutting costs. We deliver smart robotic cells for superhuman capabilities, enhancing speed, consistency,

Content Filed Under:



Seven Tips to Successfully Deploy Robots in High-Mix Manufacturing Applications

POSTED 05/23/2024

Many manufacturing companies are facing an acute shortage of workers with the right skill sets. Surface treatment processes such as sanding, polishing, coating, blasting, and spraying represent a large portion of manufacturing operations. These operations typically involve holding a tool in hand and moving it over a surface to make value-added changes to the surface. Surface treatment operations are challenging for humans because these can cause ergonomic injuries, respiratory diseases, and pose other long-term health risks. In today’s era of the gig economy, people can easily find work that does not expose them to health risks. Therefore, the labor churn rate on ergonomically challenging tasks in the manufacturing sector is very high. This is creating a major problem for manufacturing companies that cannot find people to work on surface treatment processes.

Many companies are looking for robotic automation technologies to speed up production and achieve consistent quality in surface treatment applications. Manufacturing applications can be broadly classified into two categories: (1) mass production, (2) high-mix (high-volume or low-volume).  Robots have been successfully used in mass-production applications. However, a large segment of manufacturing is high-mix and these applications pose significant challenges for traditional robotic automation. 

Unique Challenges of High-Mix Manufacturing

High-mix applications exhibit the following two characteristics. First, parts need to be changed frequently during production and the time available for accomplishing this is small (e.g., in the order of minutes). Second, there is considerable variability in part dimensions, geometry, material properties, and the part can be presented to the robotic cell. Therefore, robotics solutions that work well in mass production applications cannot be used in high-mix applications.  For example, the manual programming of robots is economically not viable in high-mix applications, because programming costs cannot be amortized over a large number of parts. This would amount to replacing workers with much more expensive robot programmers. Similarly, uncertainties in part locations cannot be reduced by using custom fixtures due to cost considerations. 

We at GrayMatter Robotics are addressing the labor shortage problem by deploying AI-powered smart robotic cells that work in high-mix applications. These solutions are improving the lives of workers in surface treatment applications by dramatically reducing the need for them to do manual work that poses health risks to them and enabling them to perform high-value work of optimizing the process and improving quality.

Robotic automation for high-mix surface treatment applications requires us  to take a fundamentally different approach compared to mass production. Components of our approach include: (1) sensor-based systems for building part models, (2) automated robot trajectory generation based on part models constructed from sensing, and (3) force-based control systems to handle part model uncertainties caused by sensor noise. In order for a robotic system to offer useful value to customers in high-mix manufacturing environments,  the system needs to have the following characteristics.  

  • Human-Competitive Quality Performance
  • Human-Competitive Cycle Time Performance    
  • Cost-Effectiveness  
  • Very High System Availability (e.g., low frequency of failures; fast recovery time when failures occur)  
  • Safety Assurance 

Common Pitfalls in Robotic Deployment in High-Mix Applications

Many companies have tried to deploy robots to manipulate tools in high-mix manufacturing applications in the past. Unfortunately, most of them saw disappointing results. Problems arise due to (1) the inability of the technology to deal with the part variability in an effective & efficient manner, or (2) long downtimes due to unscheduled maintenance and repairs, or (3) ongoing support requirements due to custom layouts or usage edge-cases, which significantly increases costs. 

Robotic cells that are deployed in high-mix surface treatment applications are complex systems consisting of multiple interacting components and operating in dynamic environments. Therefore, there is a significant potential for the onset of adverse conditions that if not handled promptly can serve as a precursor to failure. Below are a few representative examples. Pressure in the airline can fluctuate and can lead to the malfunction of pneumatic components; Suboptimal debris removal can lead to problems with imaging systems; Increased friction in the rail drive system can lead to overheating of motors; Human errors can lead to the loading of improper tools or insufficient clamping of parts. Any of these errors can lead to serious failure and cause damage. For example, if the sensing system is performing suboptimally, then it may lead to a collision that may break a cable or the tool. Recovering from serious failures requires considerable human expertise and significant downtime.  In today’s environment of supply chain delays, replacing a component in a robotic cell may take several weeks. Unfortunately, the occurrence of adverse events is inevitable in challenging environments.

Let us consider a case where after the initial careful calibration, the robot works well for a week, and then an adverse event related to degradation in sensor performance leads to a failure making the robot unavailable for ten days due to the time needed to diagnose the problem and get a replacement part. From the customer's perspective, a robotic solution that is prone to frequent failure in high-mix applications is not economically viable. Any cost savings in labor will be offset by the need to pay for expensive human expertise to diagnose the problem and perform the repair and costs associated with downtime. Therefore, we need ways to detect the onset of adverse events and proactively take actions to ensure that these events do not lead to serious failures.

Historically, the deployment of robots in mass production has avoided reliability problems by carefully controlling what enters the robotic cell and eliminating sensors by using custom jigs and fixtures. This approach mainly focuses on prevention as the strategy to reduce potential failure modes. Since the parts being processed are all identical, a system integrator can rely on extensive statistical testing to ensure reliability. This type of testing can take months and can be justified in mass production. Unfortunately, this strategy does not work in high-mix manufacturing applications due to high variability. High-mix applications require the use of sensors and real-time adaptability of tool motion and hence increase the system complexity. This leads to a much higher probability of the occurrence of adverse events.  Therefore, we cannot rely only on prevention as the strategy to avoid availability problems. We also need to utilize effective mitigation and efficient recovery as strategies to deliver good performance. 

To ensure that our customers do not get frustrated by large downtimes with our robotic cells, we have developed advanced Prognostics and Health Management (PHM) technology by deploying sensors and AI technology in our robotic cells. This allows us to monitor autonomous operation and detect the onset of adverse conditions and issue alerts. Whenever possible, alerts can trigger automated corrective actions and prevent the shutting down of the cell. When this is not possible, the cell is brought into a safe state, and an alert is sent to our remotely located support team. Our team is able to diagnose the problem using the information provided to them by sensors located in the cell and resolve the problem.

Tips for Successful Deployment

Here are seven tips for manufacturers to successfully deploy robot in high-mix manufacturing Applications:

Tip #1:

You cannot expect to purchase scanning and planning software and deploy it yourself on a robot to get the expected level of system performance. Realizing a reliable sensor-based adaptive robotic cell is challenging. Achieving a high level of system availability will require a large amount of testing that may take years to complete.  Therefore, doing this on your own is not possible unless you deploy a large number of robots to speed up the testing.  

Tip #2:

You cannot rely on a traditional system integrator to deliver high availability (e.g., 95% or higher) in high-mix applications. A system to handle part variability will require developing and deploying an AI-based PHM system to ensure that the system availability is high. Traditional system integrators build robotics systems for mass production that work well for repetitive tasks. They do not specialize in building AI-powered PHM systems.  

Tip #3:

You should look for turnkey solution providers that offer adequate support services. The system should have a built-in PHM system that can monitor the autonomous operation and alert the service team when the operation starts deviating from the expected normal operation. This will ensure that the impact of an adverse event can be mitigated. Ultimately, the combination of a system's ability to recover from errors and responsiveness of the service team will determine the availability of the system to do useful work for you.   

Tip #4:

You should look for solutions providers who can leverage scaling to deliver high availability and optimized performance at affordable costs. Collecting sufficient amounts of data from a single customer site will take a prohibitively long time to optimize the process and train the PHM. This approach is not scalable and therefore inefficient from the cost perspective. Deploying systems at multiple customers allows GrayMatter Robotics to collect sufficient amounts of data in a short period of time, fine-tune operational technology, and the PHM system. 

Tip #5:

You should use the Robot as a Service (RaaS) model. Getting support services and ensuring high availability of robotic cells is not possible under the traditional CapEx model. Under the RaaS model, customers pay a monthly or annual subscription fee to utilize the technology. This fee covers the hardware, software, training, service, and continuous updates & upgrades. The service fee is competitive with the labor savings achieved by deploying an automated robotics solution, in addition to material savings on consumables (which the robot utilizes significantly more efficiently), as well as lower rework costs due to the high reliability and repeatability of the robotic systems. Since there is no upfront capital expense, there is a low risk of adopting new technology.

Tip #6:

It is not necessary for you to deploy a system that automates 100% of the task. Often, you can realize significant benefits if you can automate even 80% of the task. This ensures that the automation solution does not become overly expensive to automate the hardest part of the job. By dividing the work between robots and humans, we can use automation today and improve the overall process.      

Tip #7:

You should look for solutions that can be used by your existing workforce. Ease of user interface and training is a key to getting a buy-in from your workforce.


To learn more about GrayMatter Robotics, visit our website. And to see our solutions in action, check out our YouTube channel.


Keep up with us:



Satyandra K. Gupta

Co-Founder & Chief Scientist at GrayMatter Robotics

Dr. Satyandra K. Gupta is the Co-Founder and Chief Scientist at GrayMatter Robotics. He holds the Smith International Professorship in the Viterbi School of Engineering at the University of Southern California and is the founding director of the USC Viterbi Center for Advanced Manufacturing. Formerly a program director for the National Robotics Initiative at the National Science Foundation, Dr. Gupta's research focuses on physics-informed artificial intelligence, computational decision-making foundations, and human-centered robotics. With over 400 technical articles, 180 invited talks, and recognition as a fellow in prominent engineering societies, he contributes significantly to the field. Dr. Gupta has also garnered media attention, featured in outlets like Economist, Forbes, Huffington Post, LA Times, and Smithsonian Magazine.