What Is Human In The Loop HITL?

hardware-in-the-loop AI optimization

The use of emulated batteries with realistic power levels helps engineers test the ability of the BMS software to cope with scenarios such as over-voltage, extreme temperature, and other fault conditions. When testing the C code for a BMS, a common approach is to use battery cell emulators as part of the hardware-in-the-loop system. Battery management systems control the operation of batteries in electric vehicles, aircraft, and energy storage systems. FPGA hardware acceleration https://aicurated.com/ai-technologies-and-tools/ai-agents-move-closer-to-hardware-aware-deployment/ generates samples at the required resolution of ~100 times faster than the closed-loop sampling frequency (plotted with MATLAB). To simulate the characteristics of these high-frequency signals, simulation environments must be capable of executing very fast sample steps, often requiring the use of FPGA hardware to achieve the required performance. Hardware-in-the-loop testing is part of the hardware/software integration step in the development workflow.

hardware-in-the-loop AI optimization

Sawyer Bowerman is an AI Developer Advocate on Red Hat’s AI team based in Boston, MA. Learn about quantization, sparsity, and advanced techniques like vLLM with Red Hat AI. Discover how to build smarter, more efficient AI inference systems. Every AI optimization you skip is a performance gain you’re leaving on the table. It’s less effective when it comes to generating content or other fresh generations. This technique works extremely well for code generation, structured output, and other predictable generation tasks.

  • “Strategic partnerships are essential, not just for bolstering defense, but for anticipating the next wave of sophisticated models and industry-specific vulnerabilities,” he told IBM Think.
  • Work will happen at both the model architecture level and in agentic systems.
  • “We’ll begin to see decentralized networks of agents that can learn from each other, share information and retain important knowledge over long horizons—weeks, months, even years,” he said.
  • “General‑purpose agents aren’t enough for legal, health or manufacturing,” Annunziata said.
  • “This ensures organizations are protected across all media formats and entry points, for all use cases, and in all toolsets—rather than relying on a single point of failure.”

For example, in the aerospace industry, hardware-in-the-loop testing of an aircraft’s flight control system using Simscape models enables engineers to simulate aerodynamic forces, hydraulic actuators, and electronic control responses. Power hardware-in-the-loop testbench components, including a power system model created in Simulink. However, some devices, such as solar inverters and electric machines, may use higher voltages and currents. In a typical HIL testbench, the HIL simulator interfaces with the controller hardware using low-current and low-voltage signals, typically up to 10 V. These voltage levels are generated and acquired from the I/O boards installed on the real-time system. This approach enables the aspects of BMS software, including battery management, cell monitoring, and power distribution units, to be thoroughly tested and validated.

Key Characteristics of HIL Optimization

  • A year ago, Matt White, Executive Director of the PyTorch Foundation, predicted that smaller models will push AI to the edge.
  • “Developers need flexible tooling for multimodal reasoning, memory components and safety-aligned evaluation, and that’s where open source thrives.”
  • “The industry validated the thesis that smaller, domain-optimized models would become central,” White recently told IBM Think.
  • While large language models remain dominant, Staar notes that the industry is hitting diminishing returns from scaling.
  • See how IBM Enterprise Advantage helps front-office teams work together to create more seamless customer experiences.
  • Since then, the open-source AI ecosystem has grown a lot, with smaller, domain-specific models achieving impressive results—it’s the case for IBM’s Granite, Ai2’s Olmo 3 and, of course, DeepSeek’s models.
  • In procurement, that means tracking requirements, spotting gaps early and suggesting fixes, giving professionals clarity and speed for fairer, faster decisions.

The connections between the real controller and the simulated plant are real analog and digital I/O. Hardware-in-the-loop simulation works by interfacing real controller hardware with a simulated physical system, known as the plant. Red Hat is an open hybrid cloud technology leader, delivering a consistent, comprehensive foundation for transformative IT and artificial intelligence (AI) applications in the enterprise. The future of enterprise virtualization for your workloads on-premise or across clouds The latest on how we reduce risks across environments and technologies The latest on IT automation for tech, teams, and environments

Quantization and Compression Validation

  • In partnership with IBM, Riyadh Air built the world’s first AI‑native airline, redefining a smarter, faster, more intuitive way to travel.
  • Led by top IBM thought leaders, the curriculum is designed to help business leaders gain the knowledge needed to prioritize the AI investments that can drive growth.
  • Discover how organizations are moving from isolated AI pilots to driving core business transformation with agentic AI.
  • IBM Granite® is a family of open, high performance and trusted AI models designed for business and optimized to scale your AI applications.
  • Purchase an individual or multi-user subscription today to access our full catalog of over 100 online courses.

Hardware-in-the-loop simulation is a technique for developing and testing embedded systems by connecting real controller hardware to a virtual environment that simulates the physical system, enabling validation of hardware-software integration before deployment. Real-time applications designed with https://aicurated.com/ Simulink can also be executed on third-party HIL systems, such as dSPACE, OPAL-RT, and NI. Simulink Real-Time, in conjunction with Speedgoat hardware, offers a robust platform for verifying and validating embedded software. Using Simulink Real-Time, you can generate and deploy real-time applications from Simulink to Speedgoat systems with the click of a button. Simulink offers an effective platform for designing and building real-time applications, making it an ideal choice for hardware-in-the-loop testing. In the automotive industry, this methodology is often referred to as rest bus simulation.

hardware-in-the-loop AI optimization

Control and software engineers can use MATLAB® and Simulink to create and execute HIL simulations. One of the key advantages of HIL simulation is the ability to test control algorithms on hardware components even before the entire physical hardware is available. By integrating real controller hardware with a virtual environment that simulates the physical system, HIL testing enables the validation of control algorithms and system interactions before deployment. Hardware-in-the-loop simulation plays an important role in the early development phases of embedded systems. This approach Artificial intelligence can include the emulation of the entire operating system of the embedded controller and communication between multiple nodes simulating service-oriented architecture (SOA) components. SIL is evolving into virtual testing, where additional controller software services are simulated.

real-world gen AI use cases from the world’s leading organizations – Google Cloud

real-world gen AI use cases from the world’s leading organizations.

Posted: Wed, 22 Apr 2026 07:00:00 GMT source

Hardware-in-the-loop testing supports hardware-software integration testing, enabling engineers to validate that the control algorithms meet all defined requirements across a variety of scenarios, including fault conditions. This testing approach evaluates every system function against its specified requirements, providing traceability and verification of intended system behavior. Real-time applications can be generated from Simulink and deployed on the real-time test systems.

Software-in-the-loop focuses on testing the compiled ChatGPT control algorithm code within a simulated environment. Both the controller and the physical plants are simulated with correspondent models. Model-in-the-loop is used at the initial stages of development to verify control algorithms using a high-level simulation model. Each technique serves a distinct purpose in the development cycle, with different fidelities of the simulation environment. Model-in-the-loop (MIL), software-in-the-loop (SIL), processor-in-the-loop (PIL), and hardware-in-the-loop (HIL) are key techniques used in the verification and validation process of embedded systems. They often include communication protocols such as UDP, TCP, CAN, and other industry-specific standards.

Automotive object detection using on real-world Sony IMX249 captures processed with the ARM Mali-C71 hardware ISP.

When it gets the prediction right, which is typically in a fraction of the time, you can effectively get 2-4 tokens for the price of 1. The draft data science model is usually a lot faster, but also a lot less accurate. You simply accept the correct tokens, reject the wrong ones, and continue onward.

Latency and Throughput Profiling

This approach involves not only simulating the plant but also incorporating real hardware that may include an entire node, comprising both the controller and the plant. In certain scenarios, hardware-in-the-loop simulation extends beyond a single control system to encompass multiple control systems or nodes. Volvo Construction Equipment Streamlines Product Development with a Real-Time, Human-in-the-Loop Simulator This process optimizes control algorithms and confirms safe operation across various flight conditions.