SPRY344A January 2022 – March 2023 AM67 , AM67A , AM68 , AM68A , AM69 , AM69A , TDA4AEN-Q1 , TDA4AH-Q1 , TDA4AL-Q1 , TDA4AP-Q1 , TDA4APE-Q1 , TDA4VE-Q1 , TDA4VEN-Q1 , TDA4VH-Q1 , TDA4VL-Q1 , TDA4VM , TDA4VM-Q1 , TDA4VP-Q1 , TDA4VPE-Q1
At a glance This white paper explains the requirements for building an efficient edge artificial intelligence (AI) system and how the vision AI processors can help optimize performance due to a heterogenous architecture and scalable AI performance.
![]() | Defining artificial intelligence at the edge. Many different kinds of
systems can benefit from edge AI processing. |
![]() | What is a practical edge AI system? Consider which architecture and cores
will best complete the tasks required of a system. |
![]() | Designing edge AI systems with vision AI processors like the TDA4 and AM6xA
systems on chip (SoCs). These SoCs are designed to deliver scalable
throughput and computing performance at low power and with lower system BOM
costs. |
When consumers order a product online, automation increases efficiency throughout every step of the process, from creating raw materials, enhancing warehouse productivity and facilitating home delivery – sometimes only hours later. Continuing these remarkable advancements in automation will require better machine perception and intelligence with fewer mistakes, which can be achieved by bringing artificial intelligence (AI) to edge devices.
Creating faster, smarter and more accurate systems requires more data from more sensors, along with increasing amounts of processing power. However, more data and computing poses challenges to a system’s performance, along with its power and cost requirements. System optimization and reduced development cycle times necessitate a practical approach to designing edge AI systems.
AI at the edge happens when AI algorithms are processed on local devices instead of in the cloud and is changing what is possible in industrial and automotive applications where deep neural networks (DNNs) are the main algorithm component. To operate efficiently in size-constrained, power and heat dissipation-constrained, and cost-constrained environments, edge AI applications require high-speed and low-power processing, along with advanced integrations unique to the application and its tasks. Figure 1 shows some of the applications where edge AI processing can be used to improve performance and efficiency. For example, edge AI systems that use vision input can implement a single camera for quality control on a production line, or multiple cameras to help support functional safety in a car or mobile robot.
Edge AI systems can help improve efficiency in warehouses and factories; make cities, construction and agriculture safer and more efficient; and make homes and retail settings smart. Let’s take a look at a few systems that require efficient edge AI processing:
Table 1 lists system requirements from various applications.
ADAS | Robotics | Smart Retail | Machine Vision |
Edge AI Box | |
---|---|---|---|---|---|
Deep learning accelerator | x | x | x | x | x |
Multicamera image signal processing (ISP) | x | x | x | x | x |
Vision accelerators | x | x | x | x | x |
Depth and motion accelerators | x | x | x | x | x |
Ethernet switch | x | x | x | ||
Peripheral Component Interconnect Express (PCIe) switch | x | x | |||
Functional safety | x | x |