Specifications
Product Family
Edge AI vision processors: TDA4VM (Jacinto 7), AM68A and AM62A
TDA4VM AI Performance
8 TOPS deep learning via the Matrix Multiply Accelerator (MMA)
TDA4VM Architecture
Heterogeneous SoC: C7x DSPs with vector AI extensions, Arm Cortex-A72, real-time MCU cores, GPU, ISP and safety hardware
TDA4VM Safety
ASIL-D functional safety for automotive and industrial robotics
TDA4VM Sensor Fusion
Multi-sensor camera, radar and LiDAR fusion on a single die
AM68A Core
Dual-core Arm Cortex-A72 vision processor with integrated deep-learning accelerators
AM68A Camera Support
Real-time processing of four to eight 2MP cameras
AM62A
Entry vision processor for one- to two-camera edge AI designs
Software Stack
TI Edge AI SDK, TI Deep Learning Library (TIDL) and the TIDLRunner bring-your-own-model flow
Model Support
TensorFlow Lite and optimized models from the TI Edge AI Model Zoo
Starter Kit
SK-TDA4VM: Linux development kit enabling 8 TOPS deep-learning performance
Starter Kit Cameras
Two CSI-2 ports plus a 40-pin connector for up to 8 cameras (with TIDA-01413 fusion card)
Starter Kit I/O
3x USB 3.0 Type-A, 1x USB 3.0 Type-C, Gigabit Ethernet, M.2 Key E and Key M, 4x CAN-FD, 4x UART
Power
Low-power operation suited to fanless and harsh-environment edge systems
Graphics
Integrated GPU and image signal processor for vision pre-processing
Applications
AMR and AGV with functional safety, machine vision stations, smart cameras, edge AI inference boxes and off-highway vehicle controls
Overview
Texas Instruments' edge AI vision processors bring deep-learning acceleration, safety-certified compute and multi-sensor processing onto single heterogeneous SoCs for machine vision, robotics and edge AI inference.
The TDA4VM, built on the Jacinto 7 architecture, integrates an 8 TOPS Matrix Multiply Accelerator, C7x DSPs with vector AI extensions, Arm Cortex-A72 application cores, real-time MCUs, a GPU, an ISP and ASIL-D safety hardware. Originally targeted at ADAS, its combination of AI throughput and functional safety also suits industrial robotics, machine vision and autonomous mobile robots.
The AM68A scales vision processing to as many as eight cameras for industrial and multi-camera designs, while the AM62A addresses one- to two-camera edge AI. The TI Edge AI SDK, TIDL and the SK-TDA4VM starter kit let developers prototype a vision-based application quickly using standard APIs such as TensorFlow Lite.
Key Benefits
Scalable, safety-aware edge AI vision. A single-vendor processor family with a shared TIDL software stack, functional-safety hardware and camera scaling from one to eight sensors simplifies design across machine vision, robotics and edge inference products.
Applications
Machine vision and smart cameras, AMR/AGV and robotics with functional safety, edge AI inference boxes, multi-camera industrial vision, smart retail and off-highway vehicle controls.
Request a Quote — TEXAS INSTRUMENTS EDGE AI PROCESSORS — JACINTO AND AM6XA VISION SOCS
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →