A Robot Dev Kit with 80 TOPS AI BPU for Real-Time Motion Control

The D-Robotics RDK S100 is a high-performance robot development kit featuring 80 TOPS of AI computing power via its Nash BPU, designed for embodied robotics applications. With its unique dual-brain architecture combining CPU, BPU, and MCU cores, it enables real-time AI inference and motion control on a single platform.

Designed for robotics and AI-driven systems, the kit integrates a 6-core Cortex-A78AE CPU, 4-core Cortex-R52+ MCU, and Mali-G78AE GPU, with 12GB LPDDR5 RAM and 64GB eMMC storage. Connectivity options include dual GbE LAN, four USB 3.0 ports, and HDMI output. There is an option for M.2 Key-M and Key-E slots for storage and wireless expansion, finally, a 40-pin GPIO header vives access to all the IOs. It supports up to 2K@60FPS HDMI displays, onboard JTAG and debug ports, and includes a development suite with 200+ open-source AI algorithms.

D-Robotics RDK S100 Specifications:

• SoC: RDK S100 Intelligent Computing Core
  • 1x Nash BPU (80 TOPS, INT8)
  • 6x Arm Cortex-A78AE @ 1.5GHz
  • 4x Arm Cortex-R52+ @ 1.2GHz
  • 1x Arm Mali-G78AE GPU @ 100 GFLOPS
• Memory: 12GB LPDDR5 (96-bit bus)
• Storage: 64GB eMMC, M.2 Key M (PCIe 3.0 x1) SSD slot
• Display: HDMI 1.4, up to 2K @ 60Hz
• Networking: 2x Gigabit Ethernet (RJ45)
• Wireless Expansion: M.2 Key E (PCIe 3.0 x1, UART) for Wi-Fi/Bluetooth
• USB:1x USB 2.0 Type-C (flashing/debug), 4x USB 3.0 Type-A
• Expansion
  • 40-pin Main GPIO header
  • 16-pin MCU GPIO header
  • JTAG, MCU expansion, camera headers (100-pin)
  • M.2 Key M (PCIe 3.0 x1) SSD slot
  • M.2 Key E For
• Power Supply: 12V–20V DC (70W/150W recommended)
• Dimensions: 120 x 121 x 51 mm (with case)

The RDK S100 includes a full development suite with over 200 open-source algorithms optimized for AI perception and robotic control. It supports heterogeneous scheduling between CPU, BPU, and MCU, enabling real-time task offloading and dynamic inference. The onboard MCU supports deterministic control tasks and high-frequency communication, while the CPU handles mid-level logic and the BPU accelerates neural network execution. More information can be found on RDK S100’s wiki page. But there is no information available on the wiki.

At the time of writing, the module is available on the Waveshare store for $479.99, and on Alibaba $475.