Application Highlight: ADLINK Robotics and Automation

June 01, 2026

Sponsored Story

Robotics are important in every industry, from healthcare and manufacturing, where they’ve been aiding human workers for many years, to cutting-edge automation solutions in warehousing, transportation, and other areas.

The pace of development is so rapid that it can be hard to keep up, but desired applications are moving even faster, outpacing even product roadmaps. This means that expert end users are looking for customizable robotics and automation platforms that can be made-to-order, as it were.

In order to reach these levels of cutting-edge robotics and keep up with the demand, Edge AI must be a component of the solution. Smart computing platforms at the Edge can power autonomous mobile robots (AMR), and quadruped and humanoid robots for any use case, and give them real-time AI inference capabilities for Edge use cases in automotive, logistics, healthcare, and service sectors, to name a few.

AMRs are the masters of indoor logistics and need to be designed for high-performance edge analytics and real-time sensor processing, along with predictive location and activity processing in dynamic environments like factory floors, warehouses, and complex facilities like hospitals and office buildings.

Quadruped robots are at the forefront of the automation game, pushing what the boundaries of robotic access and mobility can be, especially in unstructured environments like dangerous outdoor areas in mining or military applications. They need to leverage advanced edge computing and GenAI capabilities to be adaptable and flexible to changing conditions.

Humanoid bipedal robots are what many consider to be the next frontier of robotic automation, and the end goal of physical AI. Though we’re still in the early days of development and innovation, we know humanoid robots will require the most demanding computing platforms for whole-body coordination and human interactions.

Each of these three robotics categories requires a powerful compute engine and fully integrated AI processing to be effective and efficient.

Application Use Cases

AMR

Phenikaa-X was working to overcome several challenges as it worked to meet industrial requirements with its APM990 Robot and AMR Pallet Mover, deployed at the Thai Nguyen electronics factory. The solution needed to manage efficient power consumption, retain reliable operational stability, and deliver high-performance AI and image processing in a high-pressure manufacturing environment.

Robust and reliable computational performance had to be part of the solution to ensure that the robot could handle AI-driven decision-making and point cloud processing without interruptions, even in high-pressure environments, and industrial-grade reliability enabled Phenikaa-X to meet stringent operational standards consistently.

The Phenikaa-X engineers also needed to empower the APM1000 with important AI capabilities like real-time environment monitoring and obstacle detection. They required the ability to process multiple video streams simultaneously at a stable frame rate of 20 to 25 fps to operate seamlessly in dynamic factory settings. The AMR also needed to have sophisticated real-time processing capabilities to handle simultaneous sensor inputs while executing complex navigation algorithms.

Power and heat management were also key considerations because the AMR needed to be able to run for 16-hour shifts running at full power with a target of only 15W used while maintaining optimal temperatures.

The answer was found through a computing foundation that enables sophisticated real-time control of the AMR sensor arrays and actuator systems, while ensuring that safety-critical functions maintain priority even under heavy processing loads. The answer was the combination of versatile, robust, and amply validated computing elements that altogether enabled operation in both natural and artificial lighting conditions, with the ability to recognize and adapt to a range of pallet configurations common in Asian manufacturing.

Quadrupedal

Autonomous quadruped robots operating in outdoor patrol and inspection scenarios must combine advanced perception, intelligent decision-making, and precise motion control in order to ensure safe and reliable operation in all conditions. AI systems are required to fuse multi-sensor inputs, such as RGB-D cameras, LiDAR, and IMUs, while simultaneously coordinating low-latency, real-time, multi-joint leg motion across uneven and unpredictable terrain.

Developers must establish clearly defined yet tightly coordinated system layers: a perception layer that handles AI vision and sensor fusion, a deterministic control layer that manages motion and field I/O, and a rugged hardware foundation capable of withstanding dust, vibration, and wide temperature swings—all while running continuously, 24/7.

A robotics integrator developing autonomous quadruped patrol robots for hazardous environments, such as energy, chemicals, and heavy industry facilities, needs a computing architecture that can accelerate system integration, remain stable in real-world field conditions, and scale compute capability as workloads evolve while being reliable enough to avoid costly redesigns.

A developer working with quadrupedal robot applications is best served by adopting a purpose-built, dual-compute architecture combining AI perception with deterministic control and modular I/O compute.

Even in a well-designed robot, however, there are challenges. It needs to be hardened against many environmental conditions, and the compute side has to be carefully balanced. High-performance AI inference for vision workloads must not interfere with deterministic motion control. A monolithic computing approach can risk resource contention, leading to latency, jitter, and instability during motion execution – deadly in mission-critical applications.

The system requires seamless integration of RGB-D cameras, LiDAR, IMU, flexible motion control interfaces (EtherCAT or CAN), SBus remote control, and service interfaces—without relying on fragile adapters or protocol converters. Failing to address these challenges results in missed deployment windows, increased operational costs due to manual inspections, and loss of competitive advantage in the fast-moving robotics market.

In this application, the answer is in hardened, resilient hardware, combined with ruggedized and well-integrated, dedicated compute split into operational processing and AI sensing and inference.

Humanoid

At the far edge of robotics innovation are humanoid robots. One industrial humanoid solutions provider was developing a humanoid robot for use in heavy industries, where it will work alongside human teams in industrial plants, worksites, and outdoor facilities, taking on high-risk, physically demanding tasks. These robots must perceive their surroundings, plan, and act in real time, while balancing on uneven surfaces, navigating tight spaces, coordinating complex motion, and operating safely alongside people and equipment.
The answer in this use case must include a high-performance, rugged AI computing platform that fuses sensor processing, autonomous decision logic, and low-latency control while managing strict constraints on power, thermals, space, and long-term reliability.

The industrial humanoid solution provider’s unified AI model can allow humanoid robots to move confidently across diverse industrial operating environments, but there were obstacles like the need for long-term manufacturing and lifecycle stability, massive compute needs for real-time perception and control, and constrained space in a form factor that must include high-bandwidth vision, sensor fusion, and industrial control.

Overcoming these challenges required an embedded AI platform and an engineering partner built for operationally resilient robotics.

SBC35-RPL and DLAP-211-Orin Nano from ADLINK

The ADLINK SBC35-RPL serves as the processing powerhouse for all three of the above robotics applications. It ensured that the robust computational capabilities were available to meet the rigorous demands of any industrial application. The design adheres to stringent industrial standards, delivering stable and dependable performance for AMR, quadrupedal, and bipedal humanoid robotics applications. It supports audio analysis, AI-driven decision-making, point cloud processing, and any other high-impact compute needs, providing the versatility and reliability needed for these high-performance robots.

It’s powered by a 13th Gen Intel Core i7/i5/i3 Processor, dual-channel DDR5 4800MHz memory, and supports up to 64GB. It includes four independent display outputs:

• eDP/LVDS (default eDP)
• HDMI
• DP
• USB Type-C

It has SBC-FM board expansion: PCIe x1, PCIe x4, USB 3.0/2.0, LPC, and optional TPM 2.0 for enhanced security, and includes dual Ethernet ports: a 1GbE (Intel I219-V) and a 2.5GbE (Intel I225-V).

For AI inference and vision, all three applications leveraged the ADLINK DLAP-211-Orin Nano, which delivers all the required performance needed for industrial AI applications. It has low power consumption, using only 15W even during 16-hour shifts, while maintaining effective and optimal thermal management.

Stable AI-driven capabilities like obstacle detection at 20 to 25 fps ensure reliable and efficient operation in demanding environments, making the DLAP-211-Orin Nano ready for any robotic application.

Embedded with NVIDIA Jetson Orin NX/Nano, it delivers up to 100 TOPS in a compact form factor at 148mm x 120mm x 52mm and a fanless design for resiliency. It’s also prepared for extreme environments with an extended operating temperature range from -20°C to 70°C. The secret sauce for computer vision and environmental sensing is in the hardware-accelerated encode/decode for multiple HD camera streams, and includes support for multiple simultaneous deep learning models and real-time sensor fusion processing capabilities.

ADLINK’s expertise in processing, compute, and robotics makes the SBC35-RPL and DLAP-211-Orin Nano a perfect pairing for automation in industrial or any application.

Additional Resources:

• Product Page:  https://www.adlinktech.com/en/robotics
• Additional use case: https://www.adlinktech.com/en/news/adlink-partners-with-Noble-Machines