Technologies

ROS 2 has become a foundational framework for modern robotics development, offering modularity, real-time capabilities, and broad community support. However, integrating ROS-based systems into rigorous validation workflows – such as Hardware-in-the-Loop (HIL), Software-in-the-Loop (SIL), and Model-in-the-Loop (MIL) – presents unique challenges in timing, determinism, and interoperability.

This talk provides a technical overview of methodologies for embedding ROS nodes within closed-loop testing environments. Topics include synchronization of ROS communication with real-time systems, deterministic execution of control and perception algorithms, and interfacing with simulation platforms such as Gazebo, RViz, and FMI-compliant models. We will also examine the use of MCAP for scalable data logging and analysis, and discuss challenges and insights into high-fidelity data replay testing.

Through practical examples and architectural patterns, the session aims to equip robotics engineers and system integrators with strategies to validate ROS-based systems under realistic and reproducible conditions, bridging the gap between open-source development and industrial-grade testing.

AI‑ and RL‑driven control has enabled increasingly capable autonomous robots, yet real‑world deployment remains limited by the persistent sim‑to‑real gap. While most mitigation efforts focus on improving simulation or high-level control, this talk presents maxon’s complementary hardware‑first strategy. By integrating specialized firmware features and applying maxon’s deep understanding of drive‑system dynamics, key actuation and sensing nonlinearities are removed at their source. This approach delivers more predictable, simulation‑aligned behavior. Case studies from robotics partners illustrate how this approach boosts robustness, accelerates deployment, and enhances performance in real‑world autonomous systems.

As humanoids transition from research labs to real-world environments, achieving intelligence, agility, and energy efficiency requires system-level innovation across sensing, processing, power and connectivity.

This session will provide a technical roadmap for building smarter, safer, and more efficient humanoids. It will explore how integrated semiconductor technologies and optimized system design can enable the next generation of humanoids that perceive, think and move more like humans – safely and efficiently.

Drawing from real-world design examples and reference architectures, we’ll examine key design challenges including high-bandwidth sensor fusion, real-time edge AI processing, precision motor control and reliable communication between distributed subsystems. Attendees will gain insights into how advancements in embedded processing, analog signal chains and power management can reduce system latency, improve energy efficiency and increase reliability, all critical to humanoid performance and safety.

Description: AI‑ and RL‑driven control has enabled increasingly capable autonomous robots, yet real‑world deployment remains limited by the persistent sim‑to‑real gap. While most mitigation efforts focus on improving simulation or high-level control, this talk presents maxon’s complementary hardware‑first strategy. By integrating specialized firmware features and applying maxon’s deep understanding of drive‑system dynamics, key actuation and sensing nonlinearities are removed at their source. This approach delivers more predictable, simulation‑aligned behavior. Case studies from robotics partners illustrate how this approach boosts robustness, accelerates deployment, and enhances performance in real‑world autonomous systems.