Automate 2023: Flexiv Showcases Force Control in an Adaptive Robot

Applicable to polishing, deburring, machining and assembly, force control is a flourishing robotics control scheme that allows robots to regulate the force they exert on every axis and joint.

In precision robotic work, such as fitting parts into narrow gaps, minute adjustments are required that may depend on human senses. The use of components such as force and vision sensors play an active part in detecting applied force from multiple directions when robots assemble delicate parts.

“Force sensors are at the heart of Flexiv,” said Shuyun Chung, chief robotics scientist at the Santa Clara, Calif.-based robotics company. “In fact, it’s the primary reason Flexiv was created. When we founded Flexiv in 2016, force sensing tech was unreliable and inaccurate. It simply wasn’t good enough for precise robotic control.”

Shuyun Chung, chief robotics scientist at Flexiv.

At Automate 2023, the robotic automation company showcased its proprietary force sensing capabilities by demonstrating how a seven-axis adaptive industrial-grade robot called Rizon could balance a ball on a tray. The adaptive robot’s features include force control, artificial intelligence and computer vision capabilities. Detection data generated by the force sensor is what’s behind the adaptive robot’s ability to make quick adjustments when the ball encounters external disturbance, explained Connor Hughes, a lead robotics engineer at Flexiv.

“We created a force sensor that was not only more reliable but more sensitive,” explained Chung. “A good analogy for the importance of robotic force sensors are gloves. A standard collaborative robot has the same force-sensing capability as you have when you are wearing a pair of skiing gloves. A Flexiv robot is more like wearing a pair of surgical gloves. If that makes sense!”

Machine Design invited Chung to answer a few more foundational questions that help explain basic differences between force control and motion control, expound on the purpose and use of force sensors, and to help clarify their unique influence on the quality and reliability of an application.

Machine Design: What is force control, and how does it relate to robotics?

Shuyun Chung: In a very real sense, force control is how robots can “feel” their environment. Unlike traditional position-based robots that follow pre-programmed trajectories with precise movements, force control allows robots to dynamically adjust their actions based on real-time feedback from force sensors.

Thanks to force control, robots can precisely regulate the force they exert on every axis and joint, allowing for adaptable and flexible interactions with the environment. This is in stark contrast to position-based robots that rely on predefined positions. This dependence on predefined positions works well in a lab setting, but when it comes to real-world use cases where positional errors are unavoidable, position-based robots really struggle.

MD: How does it improve functionality?

SC: If a robot can “feel” what it is doing in the same way a person uses their sense of touch, it becomes far more useful, especially when it comes to delicate tasks like component assembly.

Force control basically gives the robot the ability to cope with environmental variability. For example, if a workpiece is out of alignment by 2mm, 5mm or even 20mm, it doesn’t matter. A Flexiv robot can “feel out” and understand its position.

Although humans may have a lower positional accuracy than robots, our adaptability allows us to successfully complete complicated assembly tasks. When computer vision is combined with force sensing, Flexiv’s robots can replicate this adaptability, dramatically increasing their functionality.

MD: Outline the difference between force control and motion control. Why is force control more difficult to implement than motion control?

SC: Force control and motion control are two different approaches to directing robots or robotic systems. Motion control focuses on precisely controlling a robot’s movements along a defined path or trajectory. In contrast, force control focuses on governing the forces received and applied by the robot to its environment.

Implementing force control is generally more complicated than motion control for two main reasons. Firstly, force control often requires a hybrid approach, where both motion and force must be regulated simultaneously. This adds complexity to the control algorithm and the hardware of the system.

Secondly, force control requires robots to have a very high yet precise response rate, meaning they need to quickly respond to changes in force applied to the environment. This is because the robot is already in contact with the environment during force control tasks; making even a slight error or miscalculation potentially leads to undesirable consequences.

Another challenge in force control is the effect of temperature on force sensors. Temperature changes can cause the force sensor accuracy to drift, leading to unreliable force measurements. Overcoming this challenge required us to develop our own unique force sensors, which empower our robots to be more accurate than anything else currently available.

In contrast, motion control is simpler to implement as it focuses on precise movement along a predefined path. This makes motion control more common and easier to achieve in robotic systems. Basically, motion control runs on “rails.” It follows a predetermined plan, regardless of external stimuli. Force control, though, is reactive to the environment and capable of dealing with the unpredictable nature of real life.

MD: What are the additional benefits of force control?

SC: One of the main advantages of force control is its ability to adapt to changing environments. For example, when grasping an object on a table, a position-based robot would require precise positioning information based on the table’s height and the object’s location. However, with force control, the robot can use force sensing to determine the table height and position of the object. This enables the robot to naturally adjust its movements while removing positional uncertainties and tolerating position variances.

Force control technology is particularly beneficial in the manufacturing industry, where production lines often require fine-tuning and adjustments over time. Force control has reduced the time and effort needed for production line setup, allowing for quicker and more efficient operations. It has also minimized the need for constant adjustments and retesting, as force control enables robots to adapt to changes in the production environment.

Force control brings robots a dynamic response capability, allowing for on-the-fly adjustments.

MD: What is the relationship between assembled components and force control?

SC: The relationship between assembled components and force control is crucial in ensuring precise and efficient assembly processes. Force control allows for fine-tuning robot movements based on feedback from force sensors rather than relying solely on inflexible position control.

Imagine plugging your phone charger into a wall socket in the dark. You can’t see the exact position of the socket, but you can feel it and use a gentle, waggling motion to plug it in. Similarly, in the assembly process, force control enables robots to sense the environment and make the minor adjustments necessary to fasten a screw or connect a ribbon cable.

This allows for more accurate and delicate handling of assembled components and prevents damage to the components or the surrounding environment.

Using force control in the assembly processes is like playing with Lego blocks. Flexiv’s robots can adjust their movements based on the sensed force, just like a child adjusts their hand movements when joining Lego blocks together.

Flexiv demonstrated the Rizon family of robots at Automate 2023. From easy-to-use polishing solutions to novel robotic teleoperation demonstrations, these robots are touted for their force control capacity and powerful performance.

MD: What are the main uses or applications of Flexiv’s robots?

SC: One of the primary uses for Flexiv’s robots is material removal tasks such as grinding, polishing and sanding. Our adaptive robots excel at this, especially when they are used on curved surfaces where they are required to adjust trajectory and force to match a car’s body panel or a piece of complex wooden furniture.

Force control allows our robots to automatically adjust and conform to a surface, even if it deviates from the ideal model. This simplifies the programming process as robots only need a rough trajectory, and the force control takes care of the rest. The end result is precise material removal without the need for complex trajectory planning.

Introducing force control in polishing applications has been exceptionally well-received by customers in the automotive industry, as it is seen as a highly beneficial technology that can save costs while improving efficiency. The ability of these robots to adapt to variations in the environment and dynamically adjust their applied force during tasks results in increased productivity and accuracy, making them a valuable asset to any industry.

MD: What unique design features and specifications does Flexiv bring to its robots?

SC: Flexiv brings unique design features and specifications to the manufacturing of adaptive robots through our human-inspired design ethos. We have taken inspiration from the human arm, which has seven degrees of freedom, and implemented this concept into our Rizon series of robots. This allows our robots to reach difficult positions using different postures in the same manner a human would.

What sets us apart, though, is that we control every single component in our robots. We do everything in-house, so to speak. We use a similar process to the way Apple controls its ecosystem to squeeze maximum efficiency out of every component.

Everything that has Flexiv’s name on it has been built from the ground up by Flexiv. We design, we build, we iterate, we code and we file patents.

This allows us to deliver the best robotic experience possible. Instead of using parts from multiple factories and suppliers and seeing what can be made, we decide what to build and go from there. Any new product will have teething problems, and this in-house process enables us to quickly address any issues we discover.

This level of control and customization allows us to constantly optimize our robots so they deliver superior performance. With our own force sensors, software, hardware, design, programming interface and grippers, coupled with our commitment to innovation, we are pushing the boundaries of robotic innovation.