AutoCore Launches AutoMinds™ Industrial Automation Control Software Platform for Next-Generation AI-Native Intelligent PLC/DCS Controllers

(April 1, 2026, Nanjing) – AutoCore officially launches AutoMinds™, an industrial automation control software platform. The platform systematically provides a full-stack, self-controlled software foundation for next-generation intelligent industrial control scenarios, including PLCs, DCSs, and intelligent controllers.

AutoMinds™ has already established partnerships with multiple large domestic equipment manufacturers, with key performance indicators on domestic processor platforms comprehensively exceeding design requirements, validating the platform's technical maturity in industrial control scenarios.

Industry Background: Domestic PLC Software Platforms Need a Breakthrough

The industrial control sector currently faces four structural challenges:

• Software platforms are controlled by others. The vast majority of domestic PLC manufacturers rely on overseas vendors' runtimes and programming environments, with very few self-developed logic compilation software solutions for medium-to-large PLCs. Embedded software costs account for 20%–40% of total PLC costs, and software licensing fees continue to rise, making domestic substitution imperative.
• Control systems are tightly coupled with hardware. Traditional PLC control logic, communication protocols, and underlying hardware are highly coupled — replacing a chip means rewriting the entire upper-layer software stack. With increasing supply chain uncertainties, hardware-software decoupling has become an industry imperative.
• OT/IT convergence has become an industry imperative. As Industry 4.0 advances, OT/IT convergence has shifted from trend to necessity. Traditional PLCs focus solely on real-time control, lacking native interoperability with MES, ERP, and cloud platforms, resulting in widespread data silos. The industry urgently needs a unified software platform to create an end-to-end data pipeline from the shop floor to enterprise management.
• AI-powered industrial applications are becoming the norm. AI applications such as visual quality inspection and predictive maintenance are moving from pilots to routine deployment on production lines. However, traditional controllers lack both the computing power for edge AI inference and the software framework for large model interaction. Next-generation domestic SoCs integrate NPU computing power, but lack a software foundation to safely fuse AI capabilities with real-time control.

AutoMinds™ is designed to address these challenges — not simply to replace a specific PLC programming environment, but to provide a full-stack industrial control software platform spanning from real-time kernels to communication, task scheduling, OT/IT convergence, functional safety, and application frameworks.

Core Capability 1: Layered Architecture & Broad Hardware Adaptation

AutoMinds™ adopts a four-layer decoupled architecture that completely isolates underlying hardware differences from upper-layer control logic:

• Hardware Abstraction Plane: Provides unified abstraction that shields SoC, board, and driver differences, covering mainstream peripheral interfaces including GPIO, UART, SPI, I2C, CAN, Ethernet, RTC, and Watchdog, with unified clock source, timestamp, and interrupt/IRQ configuration
• Real-time Control Plane: Carries hard real-time tasks with microsecond-level scheduling precision and deterministic execution guarantees
• General Services/Runtime Plane: Runs complete OT/IT convergence tasks and management environments, hosting AI inference, large model interaction, data acquisition, and diagnostics applications
• Device Management & Operations Plane: Handles HMI, log collection, remote diagnostics, OTA upgrades, and configuration management

The platform supports three deployment modes: standalone, dual-redundancy, and simulation/virtualization. It is compatible with mainstream hardware architectures including x86_64, ARM, and RISC-V, as well as mainstream operating systems such as bare-metal, Linux, VxWorks, and QNX. When chips are replaced, upper-layer control logic and application software require no restructuring, enabling long-term software asset reuse.

In completed projects, this architecture has been validated on domestic ARM processors through multi-gradient extreme stress testing. Hard real-time task scheduling jitter and interrupt response latency significantly outperform general industry benchmarks, maintaining stable deterministic performance even under extreme scenarios with simultaneous CPU overload, memory pressure, and I/O stress.

Core Capability 2: Multiple Real-time Levels & Execution Management

AutoMinds™ clearly defines real-time levels covering different industrial control real-time requirements:

• Sub-1ms closed-loop control tasks deployed on the dedicated hard real-time control plane with microsecond-level scheduling precision
• Periodic control and acquisition tasks with support for phase alignment and cross-plane data consistency management
• Upper-layer applications such as HMI and data analytics

The unified execution management service supports process/service start, stop, restart, dependency control, and ordered boot sequencing with a complete state machine. It also provides heartbeat, keep-alive, fault reporting, and automatic recovery strategies, implementing four-level fault recovery at the thread, process, service, and node levels.

Core Capability 3: Multi-Protocol Industrial Communication & SOA Unified Framework

Unlike traditional PLCs that support only a single fieldbus protocol, AutoMinds™ inherits AutoCore.OS's mature SOA architecture. Through a unified Communication Management abstraction layer that shields underlying protocol differences, it supports plug-in integration of multiple mainstream industrial communication protocols:

• Deterministic Ethernet: TSN protocol stack
• High-Performance Data Bus: Data Distribution Service
• Efficient Inter-Process Communication: Shared memory, zero-copy ring buffers, IPC
• Industrial IT Protocols: OPC UA Server/Client, MQTT

The platform supports control plane and information plane isolation, enabling I/O image, variable plane, and cross-plane data exchange through a unified process data plane. Taking EtherCAT as an example, AutoMinds™ deeply integrates the EtherCAT master protocol stack with the Runtime hard real-time framework, achieving 1ms-cycle hard real-time closed-loop control, with communication stability and reliability validated through long-duration continuous operation scenarios.

AutoMinds™ is not just a PLC runtime, but an industrial communication platform capable of bridging OT (Operational Technology) and IT network boundaries. Validated results show that cross-domain communication latency between the hard real-time domain and the general application domain is far below millisecond-level requirements, fully meeting the real-time data interaction needs of monitoring, data acquisition, AI collaboration, and other scenarios.

Core Capability 4: SoftPLC Runtime Deep Integration

To maximize compatibility with existing PLC ecosystems, AutoMinds™ provides deep integration capabilities with SoftPLC Runtime:

• Industrial Fieldbus: EtherCAT, Modbus, CANopen, EtherNet/IP, PROFINET, PROFIBUS, etc.
• Supports SoftPLC Runtime configuration and runtime monitoring, providing runtime status queries, log collection, and exception restart interfaces
• Supports SoftPLC Runtime data integration with the platform's persistence framework
• Supports SoftPLC Runtime and platform state machine linkage for unified lifecycle management
• For dual-redundancy scenarios, the platform supports integration with SoftPLC Redundancy SL or equivalent redundancy frameworks

This capability enables customers to retain existing SoftPLC Runtime engineering assets while gaining comprehensive enhancements from AutoMinds™ in functional safety, cybersecurity, hard real-time control, and intelligence.

Core Capability 5: Intelligent Application Domain — Native AI & Large Model Support

AutoMinds™'s dual-domain architecture provides a natural runtime environment for industrial intelligence. While the hard real-time domain ensures control determinism, the general application domain, based on the complete Linux ecosystem, natively supports the deployment and execution of AI inference and large model capabilities:

• Edge AI Inference: The platform fully leverages the NPU (Neural Processing Unit) computing power built into domestic SoCs, supporting local controller execution of AI inference tasks such as visual inspection, predictive maintenance, and process parameter optimization, enabling edge intelligence with data staying on-site.
• Safe Isolation of AI and Control: Thanks to the dual-domain architecture, AI/large model tasks run in the general application domain — even if anomalies occur, they do not affect hard real-time domain control task execution. Through cross-domain communication channels, AI analysis results are passed to the control domain as "recommendations," with control logic making final decisions, ensuring AI empowerment does not come at the cost of control safety.
• AI-Native Agents: The platform comes pre-deployed with AutoCore's AutoForge™ agent platform, evolving traditional pre-programmed controllers into intelligent devices capable of long-term upgrades and services.

Core Capability 6: Persistence, Configuration Management & High Availability

Persistence & Configuration Management: AutoMinds™ supports atomic commits, power-failure protection, versioning, checksums, and rollback, ensuring critical data is never left in a "half-written" state due to a single power loss. Boot parameters, device configurations, and application configurations are managed in layers, with audit logs and key configuration change tracking.

Reliability & High Availability: The platform provides a four-level watchdog chain — task watchdog, service watchdog, system watchdog, and hardware watchdog — supporting detection of abnormal exits, deadlocks, livelocks, resource leaks, and cycle timeouts. For high availability, it supports single-node self-recovery and dual-machine hot/warm/cold standby, providing fail-safe states and controlled degradation mechanisms, as well as crash snapshots, core dumps, and recent fault window data preservation, ensuring single-point software failures do not render the entire system unrecoverable.

Core Capability 7: Native Functional Safety & Cybersecurity

Functional Safety: AutoMinds™ inherits AutoCore.OS's standards-compliant functional safety framework, providing a three-level hierarchical monitoring mechanism — System Monitor, Health Management, and State Management. The Health Management module monitors application runtime status in real time, supporting anomaly detection and automatic recovery to ensure continuous control system availability.

Cybersecurity: The platform features a built-in defense-in-depth architecture covering the complete security chain from OS kernel to middleware to application layer:

• Application Sandbox Isolation: Each application runs in an independent space with CPU/memory quota limits
• Access Control Management (ACM): Service-level identity authentication and authorization
• Communication Security: Standards-compliant defense-in-depth communication protection with authentication, encryption, and access control; all remote management connections use encrypted channels
• Key Management: Distributed key distribution and lifecycle management (KDC) with key rotation support
• Secure Boot & OTA Signature Verification: A/B rollback support, debug interfaces can be disabled, default configurations follow the principles of least privilege and minimal service exposure

Core Capability 8: Open Interfaces & Extensibility

Internal Platform Interfaces: Provides unified service registration/discovery interfaces, unified message bus interfaces, shared memory process data interfaces, unified clock and timestamp interfaces, and unified health status, fault code, and recovery action interfaces, supporting service discovery, service registration, and interface version management.

Controller-to-Device Interfaces: Provides a unified device model that uniformly abstracts local I/O, remote I/O, and fieldbus devices, supporting plug-in integration of EtherCAT, CANopen, Modbus, EtherNet/IP, PROFINET, and PROFIBUS, as well as interface definitions for redundant links or redundant controllers.

Controller-to-Upper-Application Interfaces: Provides OPC UA Server/Client and MQTT interfaces, supporting log queries, status queries, and alarm subscriptions, with built-in authentication, authorization, and auditing capabilities.

Extensibility: The platform supports adding new bus drivers, new protocol stacks, and new runtime plugins; when adding new hardware board models, over 80% of middleware code can be reused; advanced features can be enabled via licenses or configuration.

Operations Interfaces: Provides export interfaces for logs, metrics, traces, core dumps, and packet captures; query interfaces for software inventories, version inventories, certificate inventories, and hardware capability inventories; upgrade, rollback, self-check, and health report interfaces supporting online log collection, offline diagnostic package export, and configuration snapshots.

Benchmark Project Validation

AutoMinds™'s technical capabilities have been systematically validated at partner sites. A complete industrial controller system was built on a domestic ARM processor platform, including:

• Dual-kernel real-time environment on domestic processor platform
• AutoCore.OS Runtime environment in the hard real-time domain
• Highly deterministic hard real-time domain industrial bus protocol stack
• Cross-domain communication coordination between the hard real-time domain and intelligent application domain
• Long-term system stability and high real-time performance under multi-gradient stress

All functional and performance indicators met or exceeded design requirements, surpassing industry averages.

Differentiated Advantages Summary

About AutoCore

AutoCore is an intelligent mobile computing platform product and solution provider. With the AutoCore.OS software platform at its core, the company serves the intelligent automotive, industrial automation, and intelligent robotics sectors. Its core products fully comply with the highest-level functional safety and cybersecurity standard requirements. AutoCore has received investments from renowned institutions including Bosch Ventures, Foxconn, Hillhouse Capital, and IDG Capital, and has established deep strategic partnerships with STMicroelectronics (ST), Tenstorrent, Qualcomm/Advantech, and others.