The semiconductor industry is currently undergoing a major transformation. Driven by the growing demand for computing power fueled by Artificial Intelligence (AI), it is increasingly shifting toward the development of custom chips (ASIC).
Unlike standard processors, these tailored components enable higher levels of performance, energy efficiency, and functional specialization.
A new generation of smart sensors thanks to Edge AI
One of the major advancements in this transformation is the integration of artificial intelligence directly into devices, close to the data source, without relying on the cloud: this is the principle of Edge AI.
Combined with the open RISC-V architecture and specialized hardware accelerators for AI, this approach enables the design of smart sensors that deliver:
- Reduced latency: Real-time processing
- Lower energy consumption: Fewer data transfers
- Enhanced security: Sensitive data kept locally
- Simplified systems: Reduced cloud dependency
- All while cutting costs and system complexity
This evolution shifts sensors from passive to intelligent, enabling them to understand their environment and adapt over time.
The CREA project: A French innovation, Made in Nouvelle-Aquitaine
In this context, IDMOS has won the CREA project, funded by the European Union, and is working in close collaboration with Keysom and CATIE.
Project objective: To design optimized electronic components (ASIC) tailored to specific use cases, offering better performance and significantly reducing energy consumption.
An innovative technological approach
The project is based on:
- The integration of flexible and customizable RISC-V processor cores
- The addition of dedicated hardware accelerators for AI
- The execution of algorithms directly via Edge AI, without cloud dependency
A modular and reusable approach
Starting from a standard base, the components developed under the CREA project can be configured according to a client’s specific needs, integrating their data and AI models. The goal is to simplify and accelerate the design of custom circuits while strengthening technological sovereignty and addressing the growing challenges of embedded systems and IoT.
The challenges of Edge AI design in sensors
Developing smart sensors with embedded AI involves several constraints:
- Selecting models suited to the application
- Ensuring the quality and availability of training data
- Designing dedicated hardware accelerators
- Meeting power consumption and size constraints
- Validating on FPGA before industrializing in ASIC
- Integrating into compact systems
RISC-V and Edge AI: A strategic duo
The RISC-V architecture offers numerous advantages:
- Data collection and processing from sensors
- Task management and execution of functions complementary to AI computations
- Precise adjustment of resources to the specific needs of the application
- Optimization of chip area usage and cost reduction
- Open-source and sovereign solution
On the other hand, Edge AI hardware accelerators enable:
- An optimal balance between latency, processing throughput, result accuracy, and energy consumption
- Adaptation to the characteristics of AI models and the hardware constraints of each project
Organization of the CREA project
The CREA project will run until the end of 2028.
IDMOS will handle the design and development of the ASICs.
Several engineers will be mobilized throughout the project, with additional support as needed for different development phases.
SERMA Productivity Engineering, the German subsidiary of SERMA Group, will also contribute by providing its teams for ASIC architecture and design work.
Multiple applications
This technology opens the door to numerous use cases, such as:
- Industry: Predictive maintenance, pattern recognition
- Healthcare: Wearable medical devices, event detection (arrhythmias, falls, etc.)
- Smart Agriculture: Disease detection
- Energy: Consumption analysis, fault detection
- Internet of Things (IoT): Comfort management, presence detection
Conclusion
With the CREA project, IDMOS is actively contributing to a major evolution in embedded electronics: smarter, more autonomous, and more efficient systems.
By combining custom ASICs, Edge AI, and RISC-V, this approach opens up significant opportunities for industry, healthcare, and IoT, while reinforcing European technological independence.
