Showing results for 
Search instead for 
Did you mean: 

Together with MakarenaLabs advancing Digital Twin & Robotics Technology

2 0 3,085

Digital twin innovation is a major focus for many of our customers, and Embedded World visitors were treated to an early preview of digital twin technology in action for humanoid robotics applications, courtesy of MakarenaLabs (Verona, Italy). MakarenaLabs specializes in advanced information technologies in the field of embedded cloud design, artificial intelligence, hardware acceleration and multimedia systems.

To help visualize MakarenaLabs’ value proposition in digital twin technology, imagine any system or process leveraging humanoid robotics. Now, imagine a mirror-image digital simulation of it, and consider how valuable and efficient this simulator could be for helping develop and refine the physical robot.

By leveraging virtual models of physical objects – digital twins – the synthetic data generated by sensors in the virtual model can be collected and executed in real time, and fed back to the physical robot to optimize its parameters. No more expensive, painstaking, trial and error manipulation of physical test systems and sensors to generate the needed data. Digital twin technology makes it possible to tune and optimize a robot’s performance efficiently and effectively leveraging a virtual model.

If this sounds easy, it isn’t! The hardware/software architecture is incredibly sophisticated. The FPGA and x86 CPU processing workloads are enormous.

That’s why MakarenaLabs chose AMD.

MEET PABLOphoto_2023-04-07_11-54-40.jpg

As special guests at AMD’s booth, MakarenaLabs wowed Embedded World attendees with live demonstrations featuring Pablo – a fully customizable and reconfigurable humanoid robot with moving head, eyes and mouth that’s able to detect, recognize and interact with people and objects in the surrounding environment. To perform these tasks, MakarenaLabs uses its own product, MuseBox (a real-time FPGA ML system), from which it extracts the AI models for Face Detection, Face Recognition and People Tracking.

Pablo leverages MuseBox to produce a neural network, which is then deployed onto an AMD KriaSOM to collect and synthesize data captured from the webcam in front of Pablo, as well as the onboard microphone. The demonstration system leverages an AMD Ryzen™ Embedded R2000 CPU running on the Unity framework to power the analytics and compute-intensive, real-time simulation capability, connected via high speed communication to the robot.

At the Embedded World demo, visitors used a joystick to move Pablo’s head, eyes and mouth, and were captivated as they watched the virtual replica on display. The digital twin executes a perfect mirror image of the physical model, and process optimizations are sent back to the robot for adjusting actuator positioning, voice control and more based on incoming human and environmental stimuli.

The AI within the system – built with AMD processing – makes it possible to recognize  an environment to see if there are obstacles, for example. With AI we can achieve simultaneous localization and mapping, for navigating mobile humanoid robots around obstacles via a safe path. AI is also instrumental in recognizing voices and understanding vocalized content using natural language processing (NLP) models, enabling a robot like MakarenaLabs’ Pablo to physically and/or verbally respond in context.


In the not-too-distant future when humanoid robotics technology goes mainstream for healthcare and service applications – perfected for safe interaction with humans – it will be companies like MakarenaLabs that make it possible, utilizing digital twin technology.

The powerful and seamless combination of AMD KriaSOMs and x86 CPU processing technologies will play a valuable role in enabling these AI-guided robotics applications – and many more apps on the horizon.

To learn more about the myriad embedded design benefits you can achieve with AMD Embedded solutions for industrial robotics applications, visit

About the Author
Alisha Perkins leads the marketing team for the Embedded Solutions Group at AMD. She has worked as a marketing professional in the semiconductor industry for the past 12 years.