September 18, 2026 — In a move that signals a seismic shift for the robotics industry, Aetina Corporation, a leading edge AI solution provider, has officially launched its flagship AIE-KT78 and AIE-KT68 edge AI systems. Designed to serve as the "brains" for the next generation of autonomous machines, these systems represent a critical bridge between theoretical artificial intelligence and real-world physical deployment. By integrating the formidable power of the NVIDIA Blackwell architecture via the NVIDIA Jetson Thor platform, Aetina is enabling developers to move beyond fixed-routine automation toward a future of context-aware, human-like robotics.
The Convergence of Intelligence and Motion
For years, the robotics industry has been constrained by a fundamental bottleneck: the disconnect between high-level AI reasoning and low-level physical control. Traditional edge platforms, often limited to a few hundred tera-operations per second (TOPS), have struggled to keep pace with the massive, multi-stream data requirements of modern Vision-Language-Action (VLA) models.
The AIE-KT78 and AIE-KT68 systems were purpose-built to dismantle these barriers. By facilitating the simultaneous execution of multimodal generative AI, Large Language Models (LLMs), and Vision-Language Models (VLMs), these systems allow robots to perceive their environment, reason through complex tasks, and execute precise physical actions—all locally on the device. This "on-device" capability eliminates the latency associated with cloud connectivity, ensuring that robots can operate safely and autonomously even in environments where network access is unreliable or restricted.
Technical Specifications: Power at the Edge
The architecture of these new systems is defined by their raw computational throughput and specialized industrial connectivity.
The AIE-KT78: The Performance Powerhouse
At the top of the lineup, the AIE-KT78 is engineered for the most demanding humanoid and cobot applications.
- Core Processing: Powered by the NVIDIA Jetson T5000 module.
- Memory: Equipped with 128GB of 256-bit LPDDR5X memory, ensuring rapid data access for massive model weights.
- AI Throughput: Delivers up to 2,070 FP4 TFLOPS, providing the necessary headroom for complex neural network inference.
The AIE-KT68: Efficiency and Versatility
The AIE-KT68 offers a more balanced profile for projects where power consumption and system cost are critical variables.
- Core Processing: Powered by the NVIDIA Jetson T4000 module.
- Memory: 64GB of 256-bit LPDDR5X memory.
- AI Throughput: Delivers up to 1,200 FP4 TFLOPS.
Both systems are housed in a remarkably compact form factor—just 80mm thick—making them suitable for integration into tight robotic chassis where space is at a premium. They support a wide power input range (9–48VDC) and are designed to function reliably in extreme temperature conditions ranging from -25°C to +55°C, ensuring durability in factory floors, warehouses, and outdoor environments.
Chronology of Development: From Concept to Production
The release of the AIE-KT series follows a multi-year trajectory of rapid evolution in the edge computing sector.
- 2024–2025: The industry saw an explosion in the development of Vision-Language-Action (VLA) models. However, these models were largely trapped in simulation or server-grade hardware, proving too heavy for edge deployment.
- Early 2026: NVIDIA’s unveiling of the Blackwell-based Jetson Thor architecture provided the necessary "compute-per-watt" shift required for embodied AI.
- Q2 2026: Aetina began close collaboration with developers to integrate the Thor platform into a ruggedized, industrial-grade chassis.
- September 2026: General availability of the AIE-KT78 and AIE-KT68, marking the transition from experimental R&D to mass-production-ready hardware for global roboticists.
Sensor Fusion and Deterministic Control: The "Nervous System"
One of the most significant innovations in the AIE-KT series is its ability to act as a unified nervous system for a robot.
High-Bandwidth Perception
To make sense of the world, a robot requires constant data flow. The AIE-KT series utilizes QSFP28 connectivity, capable of delivering up to 4x25Gbps of bandwidth. This allows for the integration of high-resolution cameras, 3D LiDAR, radar, depth sensors, and IMUs. With support for up to 8 GMSL2 camera inputs, the systems provide the high-fidelity spatial understanding required for humanoid navigation and complex object manipulation.
The EtherCAT Advantage
Perhaps the most crucial feature for industrial utility is the dedicated RJ45 1GbE EtherCAT port. Acting as an independent EtherCAT Master, this port allows for microsecond-level synchronization between the AI’s "decision" and the hardware’s "action."
- For Cobots: This means the ability to safely switch tasks and adapt to human presence in real-time.
- For Humanoids: It enables continuous, fluid motion, preventing the "stutter" often seen when compute and motion controllers are decoupled.
By consolidating these functions, Aetina eliminates the need for redundant control hardware, thereby reducing the complexity of the robot’s internal architecture and lowering the total cost of ownership.
Official Perspectives: The Industry Shift
Richard Hung, Vice President of Product Division at Aetina, emphasized that the competitive landscape for robotics has moved beyond raw speed.
"The competitive edge for cobots and humanoid robots has shifted from the compute performance of a single model to whether a system can integrate perception, reasoning, decision-making, and action in real time in real-world environments," said Hung.
He highlighted that Aetina’s value proposition extends far beyond the hardware itself. "As an NVIDIA Elite Partner, we combine over a decade of edge AI engineering experience with a fully validated peripheral ecosystem. We provide end-to-end technical support—from mechanical design and thermal engineering to BSP and system integration—to help developers and system integrators reduce integration risk and accelerate the path from proof-of-concept to scaled deployment."
Implications for the Robotics Ecosystem
The arrival of the AIE-KT78/68 series has significant implications for several sectors:
- Industrial Automation: Factories can now deploy robots that are not just pre-programmed to follow a path but can interpret visual changes in the production line and adjust their movements accordingly without human intervention.
- Human-Robot Collaboration: The deterministic control capabilities allow for safer co-working environments. Robots can react to human movement with the same speed and intuition that a human would, drastically reducing safety risks.
- Humanoid Research and Development: By providing a standardized, high-performance platform, Aetina is likely to accelerate the development cycle of humanoid robots, which have historically been hindered by the difficulty of managing internal sensor fusion and motion control.
- Software Integration: Because these systems support the full NVIDIA AI software stack—including JetPack 7, CUDA, TensorRT, and Isaac ROS—developers can leverage a massive library of existing tools. This software parity ensures that code developed in simulation environments can be ported to the physical robot with minimal friction.
Conclusion: The Path Forward
As industries scramble to address labor shortages and the need for higher operational efficiency, the AIE-KT78 and AIE-KT68 appear at a pivotal moment. By solving the dual challenges of massive AI compute requirements and precise physical control, Aetina is not merely selling a product; they are providing the foundational architecture for the next era of automation.
For developers, integrators, and businesses, the availability of these systems signifies that the era of "smart" machines—those that can truly perceive, reason, and act in the physical world—has officially moved from the laboratory to the factory floor.
For more detailed technical documentation or to inquire about procurement, interested parties can visit the official Aetina product page for the AIE-KT78 and AIE-KT68.
