Force feedback machining

Big Data! The Sensor Revolution! IoT! Industry 4.0! These terms are flying about with promise of improvements in capacity and efficiency, ushering in the “Third Industrial Revolution!” There is always hype when talking about what new technologies can deliver, but I see real and significant potential advances and will come to those who have the boldness, vision, and technical talent to effectively and efficiently develop these new technologies.

Let’s focus on one of the ways that these new technologies could impact an old technology: machining.

Today’s machine tools are incredibly sophisticated pieces of technology that showcase the culmination of countless engineers’ ground-breaking work to make machine tools faster, with higher accuracy, lower cost, and more sophisticated controls. However, machine tool controls are still largely based in the relatively narrow confines of position and velocity feedback control, which is fundamentally limiting. Advances in sensing capability, sensor miniaturisation, signal processing, algorithm design, and machine learning that are happening right now will enable a step change in manufacturing processes and strategies.

This is an expansive topic, so I’ll just look at one area of impact: force sensing and force feedback control. First, I will look at how force sensing and feedback control has the potential to enhance machining processes (or any material removal process that relies on mechanical interaction). Second, I will look at the key challenges to implementing this technology.

Force feedback is a hot topic in robotics and some have even made initial attempts to apply it to machining.

The benefits

So, what are some of the benefits to integrating force sensing into a machine tool?

• Faster cycle times – With force feedback, a CNC could automatically increase the feed rate when the machining forces are small to maintain an effectively constant chip load (likely only for roughing operations). There are software packages available that will optimize reduce cycle time based on complex physical or analytical models, but these estimates are only as good as the inputs and algorithms. Machining is a very complex, highly non-linear thermo-mechanical-chemical process with many changing variables, and these models cannot truly represent the machining process forces. But by measuring the forces in real time, this data can be used to adapt the feed rate and spindle speed, allowing the machine to speed up in regions where the cutting forces are too low (wasting machine potential) or too high (putting undue wear on the machine tool bearings, motors, and ways).
• Increased tool life – Measuring the cutting forces in real time would enable the ability to reduce the feed rate when the machining forces are too large, preventing excessive tool wear or tool breakage.
• Increased yield – Measuring the cutting forces in real time and comparing to historical cutting force data would enable the ability to alert when tools are dull, preventing warped parts from increased machining induced stresses or part damage around delicate sections (e.g. thin-wall areas).
• Prevent crashes and/or minimize damage – A force sensing machining center with sufficiently high bandwidth could sense force spikes and prevent the system from crashing due to programming error or unanticipated objects in the machine path.
• Prevent chatter – By performing a frequency analysis on the data, the system could detect chatter and either adjust the feed or spindle speed.
• Make machines safer for people to work with – By setting force thresholds very low during jogging operations, a CNC could stop immediately if the system runs into a person or something else.

The opportunities

Second, why hasn’t it been done already? What are the major challenges? The fundamental sensing technology is mature enough, so why hasn’t this been done already? Of course, there are many obstacles associated with integrating force sensing into machine tools. There are foreseeable challenges and challenges that only emerge during the development. Some of the foreseeable challenges are:

1. High resolution, medium-high frequency sensing in a challenging environment for a reasonable cost. A force sensor that performed all of the tasks listed above would need a sensing range from newtons to kilonewtons for most machining applications (clearly micro-machining operations and very large machining operations would have different force sensing requirements). The sensors would also need to be able to sense up to a kilohertz (maybe even up to 3 kilohertz, depending on the application). And although these requirements aren’t necessarily impossible, doing them inexpensively and robustly is not a trivial challenge.
2. High speed wireless communication in a very noisy environment – Ideally, it would be best to measure the forces as close to the action as possible, and given the complex and constantly changing force vectors in machining, mounting the force sensor(s) on the tool holder seems like the best option, except for the pesky annoyance that it’s rotating very quickly and that it’s constantly being changed in and out of the spindle. This would make any wired connections (even to a slip ring), unfeasible. The slip ring would be noisy and would likely wear out quickly in such a harsh environment. But we are living in the future and high-speed wireless communication is a thing! The challenges come from the fact that there might be tens to hundreds of machines in the same facility, which makes wireless communication quite complex, not even considering the EM noise from the motors. Further complicating matters is that the sensors and wireless communication would likely need to consume low power, unless a clever way of passing them power were devised.
3. High speed computation and data processing – Once the data comes out of the machine, it will be necessary to perform some clever algorithms on it to get any meaningful information that we can use to tell the machine to speed up, slow down, stop, etc. 
4. Sophisticated control algorithms that combine position and velocity with forces like impedance control, admittance control, hybrid control architectures, or others.
5. Sophisticated control strategies to optimize for one or more parameters (e.g. faster cycle time, less tool wear and part deformation, and less chatter).
6. Integration of a complex system for robust performance.

None of these problems individually is in the realm of science fiction. Each has been solved to some extent in other fields. Every day people around the world are making advances in sensor miniaturisation, signal processing, algorithm design, and machine learning.

Force feedback is just a subset of the broad group of technologies that are opening up new possibilities in manufacturing. These technologies will have a drastic impact on machining technology and manufacturing as a whole. And I can’t wait to see it happen.