Sensor fusion

Meet ever-increasing demands to compute sensor and vision data in vehicles quickly, accurately and reliably

As vehicles reach unprecedented numbers of sensors, there is a need for sensor fusion systems to interpret data and make split-second driving decisions. By transmitting high-resolution, synchronized sensor, camera and radar data and processing it in real time, our power-efficient, high-performance processing and communications technologies help enhance overall vehicle safety.

Why choose TI for your sensor fusion systems?

checkmark

High-performance processing

Create more accurate and reliable perception using highly-efficient vision and  artificial intelligence (AI) accelerators to combine and process data from multiple sensors and sensor modalities.

checkmark

Reliable, synchronized communication

Transmit and synchronize real-time data about a vehicle's environment for accurate, intelligent processing. 

checkmark

Easy system implementation

Start designing and get to market faster with our evaluation boards, compatible software and application-specific examples.

Design the sensor fusion platform for the autonomy level you seek

Combine and process sensor data in real time

Advanced driver assistance systems require the vehicle to make safety decisions in real time. From computer vision and signal and image processing to artificial intelligence (AI) and deep learning capabilities, our Arm®-based processors enable a broad range of efficient computing performance to improve automotive safety and enable the next step in vehicle autonomy. Design high-performance sensor fusion systems without sacrificing critical resources that streamline system design while helping meet ISO26262 standards up to ASIL D.

Featured products for real-time processing
TDA4VM-Q1 ACTIVE Automotive system-on-a-chip for L2, L3 and near-field analytic systems using deep learning
NEW AM2634-Q1 ACTIVE Automotive quad-core Arm® Cortex®-R5F MCU up to 400 MHz with real-time control and security

Reliable and secure data transmission across various protocols

As the number of cameras and sensors increase in the vision of building a fully autonomous vehicle, there is more data from around the vehicle than ever. Our advanced communication technologies safely and reliably transmit sensor data across any vehicle architecture, enabling time-critical safety decisions. Our vehicle networking portfolio includes Ethernet PHYs, FPD-Link devices, CAN, LIN, and PCIe supporting various data speeds and improving reliability for safety-critical driver assistance applications.

Featured products for multiprotocol communications
DP83TG720S-Q1 ACTIVE 1000BASE-T1 automotive Ethernet PHY with RGMII & SGMII
DS90UB960-Q1 ACTIVE Quad 2 MP camera hub FPD-Link III deserializer with dual CSI-2 output ports
TCAN1043A-Q1 ACTIVE Automotive CAN FD transceiver with wake/inhibit/sleep

Power-efficient performance

Reducing power needs while meeting growing performance requirements is a concern for designers of advanced driver assistance systems. A discrete integrated power-management integrated circuit (PMIC) dedicated to a specific processor system-on-chip (SoC) helps reduce component count and system size while increasing efficiency and thermal performance. Our PMICs help you limit noise and maintain power density in sensor fusion designs while helping meet ISO26262 standards up to ASIL D.

Featured products for power efficiency
TPS6594-Q1 ACTIVE Automotive 2.8-V to 5.5-V PMIC with five buck regulators and four low-dropout regulators
LP8764-Q1 ACTIVE 4 5-A/20-A multiphase buck converters PMIC for automotive SoCs
NEW TPS62870-Q1 ACTIVE Automotive 2.7-V to 6-V input, 6-A, stackable, synchronous buck converter

Accelerate software development on TDA4x processors

Evaluate software performance on TDA4x processors on our evaluation modules (EVMs) and in our cloud environment. Easily compile and deploy your models and accelerate inference with industry-standard application programming interfaces such as TensorFlow Lite, Open Neural Network Exchange (ONNX) Runtime, Tensor Virtual Machine (TVM), GStreamer, Docker, Robot Operating System (ROS) and Open Graphics Library (GL) for Embedded Systems (ES) to accelerate deep learning inference.

Featured products for algorithm development
TDA4VM-Q1 ACTIVE Automotive system-on-a-chip for L2, L3 and near-field analytic systems using deep learning

Optimize sensor fusion performance with our clocks and timing devices

Clocking is vital in sensor fusion to deploy multiple high-performance performance SoCs. Our portfolio of low-jitter, easy-to-use clocks and timing devices, including bulk acoustic wave oscillators, allows you to build your clock tree with simple discrete devices or highly integrated solutions.

Featured products for reliable clocks & timing
CDCS503-Q1 ACTIVE Automotive clock buffer/clock multiplier with optional SSC
CDCE913-Q1 ACTIVE Automotive catalog programmable 1-PLL VCXO clock synthesizer with 2.5-V or 3.3-V LVCMOS outputs
LMK00334-Q1 ACTIVE Automotive 4-output PCIe® Gen1/Gen2/Gen3/Gen4/Gen5 clock buffer and level translator

Design & development resources

Reference design
ADAS 8-Channel Sensor Fusion Hub Reference Design with Two 4-Gbps Quad Deserializers
This sensor fusion hub reference design allows the connection of up to four 2-megapixel cameras and up to four radar modules over coaxial cable. This design utilizes these coaxial cables to provide power, backchannel communication, and clock synchronization to the sensors. The two 4 Gbps FPD-Link (...)
Evaluation board
TDA4VM processor starter kit for edge AI vision systems

Bring smart cameras, robots and intelligent machines to life with the TDA4VM processor starter kit. With a fast setup process and an assortment of foundational demos and tutorials, you can start prototyping a vision-based application in less than an hour. The kit enables 8 trillion operations per (...)

Reference design
Automotive domain controller for gateway, assisted and automated driving systems reference design

Decentralized vehicle architectures on the road today use individual ECUs that lack processing power and high-speed interfaces to handle the complex tasks and data movement needs of newly emerging automotive architectures. Higher level functions require the right combination of DMIPS, data (...)

Reference designs related to Sensor fusion

Use our reference design selection tool to find designs that best match your application and parameters.

Featured products

Arm-based processors TDA4VM-Q1 ACTIVE Automotive system-on-a-chip for L2, L3 and near-field analytic systems using deep learning
Multi-channel ICs (PMICs) TPS6594-Q1 ACTIVE Automotive 2.8-V to 5.5-V PMIC with five buck regulators and four low-dropout regulators
Arm Cortex-R MCUs AM2732-Q1 ACTIVE Automotive dual-core Arm® Cortex-R5F MCU up to 400 MHz with C66x DSP, Ethernet, safety, security
Ethernet PHYs DP83TG720S-Q1 ACTIVE 1000BASE-T1 automotive Ethernet PHY with RGMII & SGMII
FPD-Link SerDes DS90UB960-Q1 ACTIVE Quad 2 MP camera hub FPD-Link III deserializer with dual CSI-2 output ports
CAN & LIN transceivers & SBCs TCAN1043A-Q1 ACTIVE Automotive CAN FD transceiver with wake/inhibit/sleep
07 Jun 2022 | COMPANY BLOG

As ADAS technology extends to critical, time-sensitive applications – such as emergency braking, front-collision warning and avoidance, and blind-spot detection – combining data from multiple sensors enables reliable, real-time decisions for safer autonomous driving.

Read more

Technical resources

More literature
More literature
Paving the way to self-driving cars with ADAS (Rev. A)
Read this document  to understand how we are paving the way to self-driving cars with ADAS.
document-pdfAcrobat PDF
More literature
More literature
Efficient Visual Localization on TDA4VM (Rev. A)
Read this document to learn about efficient visual localization on TDA4VM.
document-pdfAcrobat PDF
More literature
More literature
Communication Protocols in Modern ADAS Architectures
This white paper examines the communication technologies enabling reliable advanced driver assistance systems.
document-pdfAcrobat PDF