Analog Tip Design of Functionally Safe, AI-Enabled Robotic Systems

Related Vendors

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

Robotic systems are becoming increasingly complex. Real-time capability, security and functional safety are fundamental requirements. Which processors, sensors and sensor interfaces are suitable?

Robotic systems now require intelligent, safe and real-time capable processing platforms. Furthermore, functional safety is often one of the fundamental requirements, especially if the planned systems are to be used in close proximity to people.

The processors used must therefore be selected with this in mind and have diagnostic functions such as voltage and temperature monitors, ECC-protected SRAM and separate power supply and clock areas. FFI (Freedom from Interference) is another requirement that enables the safety mechanisms to remain active even if a failure occurs in the application area.

Certifications such as IEC 61508 SIL-3 or ISO 26262 ASIL-D attest to the suitability of the components for safety-critical industrial and autonomous robots. Figure 1 provides an overview.

Robot systems must also enable a high-throughput and low-latency fusion of different sensor modalities so that program functions can be executed safely in real-time environments. Multiple cores and hardware accelerators are often required to handle the various data processing operations.

SoCs with integrated support for sensor interfaces such as CSI-2, CAN-FD or Ethernet as well as internal memory and high-speed connectivity enable efficient data transfer and merging. Overall, this results in the necessary real-time performance for mission-critical control loops, while demanding AI and vision tasks are outsourced to dedicated subsystems.

Intelligent Solutions Directly at the Edge

The Edge AI Studio tool (Figure 2) simplifies model development, benchmarking and deployment, while the TI Model Zoo provides a range of pre-optimized networks such as YOLO, MobileNet and ResNet for retraining with application-specific data. This allows robotics developers to rapidly iterate and deploy intelligent solutions directly at the edge.

Security is another essential pillar for robot systems, especially when working together with humans, with a cloud connection or when processing confidential data. In addition to a secure boot process, the processors used must have a hardware root of trust and memory protection units that support secure data processing and firmware validation. Arm TrustZone also ensures the distinction between secure and non-secure processes.

Highly Integrated, Safety-Optimized SoC

The processors in the TDA4 family meet these challenges with a highly integrated, safety-optimized SoC architecture. With their heterogeneous multicore structure, they are specially designed for robotics applications. Powerful Arm Cortex-A72 or -A53 processors for the operating system and applications are supplemented by real-time-capable Arm Cortex-R5F cores in a dedicated, isolated MCU island.

In addition, there are hardware accelerators for imaging, image processing and deep learning inference. Security-critical operations can thus be isolated from general tasks, making design easier. Functions such as Secure Boot, TEE (Trusted Execution Environment) and secure debug support provide customers with the assurance that their products can reliably perform the tasks assigned to them.

There is a comprehensive, standardized software development environment for the TDA4 family - including support for open source and real-time operating systems such as FreeRTOS and Linux. The TI Software Diagnostic Library is also available for safety mechanisms.

The TDA4VM is used in practice, for example, in Amazon's autonomous mobile robot "Proteus", which is used safely alongside humans in warehouse environments. (kr)

*Dhane Jones is System Engineer Robotics at Texas Instruments in Dallas / USA. Saravan Narayanan works in Product Marketing Embedded Processors at Texas Instruments in Dallas.