Processors

TDAx ADAS SoCs – Front camera

Front camera systems primarily rely on one or more camera sensors for perception. While cameras are the most precise mechanism available today to capture accurate data at a high resolution, typical ADAS system implementations use multiple sensors and modalities to enable more reliable scene perception based algorithms and achieve higher levels of Automotive Safety (ASIL). Typically, an automotive front camera integrates with another RADAR sensor ECU to enable sensor fusion, for higher reliability of perception and processing, required by functional safety targets. These systems can scale from being informational to enabling critical active safety controls as well as designing complex autonomous systems.

Many of these applications are required to meet the standards for new car safety assessment programs mandated by safety authorities such as European New Car Assessment Program (Euro NCAP) and the US National Highway Traffic Safety Administration (NHTSA). In order to make these applications affordable in entry- and mid-level vehicles, it is important to have a cost-effective, yet high-performance and a low-power solution.

Mono front camera

Mono front camera

  • VGA to 8 MP
  • Multi ROI processing
  • >16 FPS processing
Stereo front camera

Stereo front camera

  • VGA to 3 MP
  • Multi ROI processing
  • >22 FPS processing
Multi-optic front camera

Multi-optic front camera

  • VGA to 3 MP
  • Multi ROI processing
  • >22 FPS processing
Automotive Processors Front Camera Block Diagram

Features and benefits

Scalable platform

TI Jacinto™ TDAx automotive processors enables customers to

  • Completely own software & hardware implementation
  • Scale between mono and multi-camera use cases as well scale across multiple camera resolutions
  • Differentiate with features and performance, while enabling ease of development using open compute frameworks 

Heterogeneous architecture

Heterogeneous and scalable architecture provides the optimal mix of performance, low power and safety, thus optimizing the overall system cost

Functional safety

  • Systems capable of achieving ASIL-B to ASIL-D safety levels
  • Various safety OS options
  • AUTOSAR MCAL (Microcontroller Abstraction Layer) solutions

Development support

TI Jacinto TDAx automotive processors are supported by a mature software, hardware,  and 3rd party ecosystem:

  • “Easy to Develop” software frameworks with support for Open Frameworks like OpenVX/CL
  • Processor Software Development Kit for Vision Applications including specific use cases for front camera algorithms
  • Multiple cost effective hardware development platforms for every stage of development
  • Rich 3rd party ecosystem offering various levels of support from proof of concept to production including production ready algorithms with rich details on KPIs 

Automotive quality

  • Purpose built for automotive with strict quality and reliability controls and SafeTI™ process
  • Reliable production fabs and long-term availability guaranteed

Product family selection guide

Feature Set
TDA3x (TDA3xD)
TDA3x (TDA3xR)
TDA2x (TDA2xD)
TDA2Px (TDA2PxR)
TDA Device Scalability Pin2Pin compatible Pin2Pin compatible
Front Camera Feature Set
no
no
Cross Traffic Alert (CTA) Optional ISP Integration
no
no
Light Source Recognition
no
Traffic Jam Assist (TJA) Standard feature scaled up in performance
no
Cyclist Detection
Pedestrian Detection Standard feature scaled up in performance, resolution support
Traffic Sign Assist (TSR)
Lane Keep Assist (LKA)
Up to 1.3MP Up to 1.3MP Up to 2.5MP Up to 2.5MP
1x Camera 1x Camera 3x Camera 3x Camera
MPU (DMIPS) Up to 0.5K Up to 0.5K Up to 10.6K Up to 12.7K
EVE 1xEVE 800 MHz 1xEVE 1 GHz 4xEVE 800 MHz 2xEVE 1 GHz
C66 DSP 2xC66 750 MHz
2xC66 1 GHz 2xC66 750 MHz 2xC66 1 GHz
Memory BW (GB/s) Up to 4.0 Up to 5.3 Up to 8.5 Up to 10.7

Hardware and software platforms

TDA3x Evaluation Module Kit

General purpose evaluation module based on TDA3x processor that integrates a host of peripherals including multi-camera interfaces (both parallel and serial), CAN-FD, Ethernet PHY, display, SERDES module and support for interfacing with Safety MCU EVM.

DesignCore™ TDA3x Automotive Starter Kit

This benchtop starter kit lets you evaluate embedded ADAS technology. Based on the TDA3x advanced vision processor, it enables synchronous acquisition of four FPD-Link III HD data streams (video, radar, lidar, etc.) for real-time vision processing and analytics.

TDA2x Evaluation Module Kit

General purpose evaluation module based on TDA2x processor. The board supports interfacing with vision application board and 1920x1200 multi-touch display that will help jump start your evaluation and application development.

Automotive Vision Software Development Kit (SDK)

TI's innovative vision SDK framework allows developers to create different ADAS application data flows involving video capture, pre-processing, analytics algorithms and display. TI's vision SDK is complemented with libraries for EVE and DSP cores, enabling users to jump start development.

Software components and applications 

Software components
Front camera applications
SOC TDA3x, TDA2x, TDA2Px
Boot loader, peripheral drivers SBL, drivers for standard peripherals (I2C, SPI, etc.)
Capture driver CSI2 or VIP drivers for multi-channel Image capture
ISS drivers
Drivers for Image pipe, HW de-warp
Image tuning tool
PC tool for image quality tuning
Video codecs
(MJPEG / MPEG4 / H.264)
Hardware / software video encode for recording / Ethernet display
Compute library
Compute library on EVE and DSP
TIDL library for deep learning
Front camera algorithms
Demo algorithms for pedestrian detection, traffic sign recognition, traffic light recognition, lane detection, front collision waring, etc.
Links framework
Framework for connecting drivers, algorithms and other processing nodes
Demo applications
Front camera demo
AUTOSAR MCAL drivers
MCAL drivers for AUTOSAR
Security
SECDEV package for boot / run time security with HS devices

Find support for all TDAx based TI hardware platforms, tools and software offerings.

TI's innovative vision SDK framework allows developers to create different ADAS application data flows involving video capture, pre-processing, analytics algorithms and display.