EtherCAT and Real-Time Software for Humanoid Robotics Applications

Coordinated Multi-Axis Motion Control

Humanoid robots feature numerous servo-controlled joints across the spine, arms, legs, and hands, often exceeding 30–40 synchronized motion axes. Achieving lifelike movement requires sub-millisecond precision and full deterministic coordination between drives.

acontis’ EC-Master EtherCAT MainDevice (EtherCAT Master) Stack enables ultra-fast communication across all motion components with minimal jitter. By leveraging advanced features such as Distributed Clocks (DC) and Split Frame Processing, EC-Master allows developers to achieve highly synchronized, smooth joint trajectories even in complex full-body motion scenarios. Furthermore, acontis offers Optimized Real-time Ethernet Drivers that dramatically improve the performance and reliability of EtherCAT implementations.

With acontis’ EC-Motion EtherCAT Motion Control library, developers can control CiA402®-based servo drives without needing additional hardware. EC-Motion provides a PLC-style interface and supports both single-axis and multi-axis movement. It includes a built-in trajectory generator that calculates smooth, jerk-limited motion profiles based on defined dynamics such as speed, acceleration, deceleration, and jerk. Drives can be operated in either Cyclic Synchronous Position (CSP) or Cyclic Synchronous Velocity (CSV) mode, making EC-Motion ideal for full-body coordination in humanoid robots.

Sensor Fusion, Feedback Loops, and Safety

Humanoids rely on real-time integration of diverse sensor types, including IMUs, torque sensors, joint encoders, LiDAR, and microphones. These sensors must be polled, fused, and responded to in a single control cycle for proper balance and interaction.

acontis EC-Master supports high-frequency feedback loops and precise timestamping across distributed devices, making real-time sensor fusion and reactive control possible. Additionally, EC-Monitor provides passive network analysis for verifying timing constraints and identifying anomalies without interfering with runtime performance.

To meet the growing processing demands of AI and real-time sensor data, acontis has teamed up with NVIDIA to bring optimized EtherCAT performance to Jetson platforms like Orin and Thor. This pairing gives humanoid robots the power to analyze vision, audio, and movement data in real-time while still maintaining the precise and deterministic control required for EtherCAT-based motion. It’s a powerful combination that brings perception and action together on a single, unified platform.

Safety is a growing concern in humanoid robotics, particularly when operating around humans. Even in a safe state, communication between critical devices must be maintained reliably and securely. The EC-Master stack is a strong foundation for implementing safety protocols. It supports the black channel principle, ensuring compatibility with FailSafe over EtherCAT (FSoE), and enables continued EtherCAT communication even during safety events. Developers can handle safety-critical messaging alongside regular control traffic with confidence.

Deterministic Real-Time Communication

Maintaining dynamic balance, executing collision-free motion, and adapting to external stimuli all require guaranteed, low-latency communication between the MainDevice application and distributed drives, sensors, and I/O.

acontis’ EtherCAT MainDevice stack delivers hard real-time responsiveness, ensuring that command and feedback cycles are completed with microsecond-level determinism. This minimizes jitter, delays, and unpredictability that could otherwise lead to instability or failure in motion-critical tasks such as walking, climbing stairs, or reacting to disturbances.

For systems with multiple concurrent control loops (e.g., limb kinematics, balance control, and sensor polling), acontis’ software enables developers to structure deterministic communication architectures that scale across all sub-systems in a full humanoid robot.

Developers can now integrate EtherCAT-based real-time communication seamlessly into the ROS 2 ecosystem—combining deterministic fieldbus control with the flexibility and modularity of ROS 2. This allows humanoid robots to synchronize high-frequency motion control with higher-level AI, vision, and behavior-planning nodes, all within a unified, standards-based framework.

Hardware Resource Constraint

Space, power, and heat are critical constraints in humanoid platforms—particularly those running on battery power. Developers need lightweight control software that doesn’t compete with AI workloads for compute resources.

EC-Master has been tested and optimized for leading embedded platforms, including NVIDIA Jetson and Qualcomm robotics processors. The software is highly efficient, generating minimal CPU load, keeping thermal output low, and preserving more than 99% of CPU availability for mission-critical applications like AI inference, perception, or navigation. This improves key priorities for untethered humanoid designs, such as battery life and system responsiveness.

Modular and Hot-Pluggable End Effectors

Humanoid robots may be designed with interchangeable hands, tools, or sensor modules depending on the application. Supporting this flexibility requires the ability to hot-connect or swap devices without rebooting the system or rebuilding the EtherCAT network configuration.

With acontis’ Hot Connect feature pack for EC-Master, humanoid developers can design modular architectures that support dynamic device attachment and removal. This allows new end effectors, tools, or subsystems to be added on the fly, whether for development testing or field-level task changes, without sacrificing uptime or real-time performance.

acontis also offers a powerful EtherCAT Simulation library, EC-Simulator, that can be utilized to test the EtherCAT MainDevice application. For example, EC-Simulator can be used to force SubDevice failure, cable breaks, and countless more error states, many of which are difficult or impossible to replicate in physical systems.