nxp edgeready smart touchless hmi solution
TRANSCRIPT
PUBLIC
NXP, THE NXP LOGO AND NXP SECURE CONNECTIONS FOR A SMARTER WORLD ARE TRADEMARKS OF NXP B.V.
ALL OTHER PRODUCT OR SERVICE NAMES ARE THE PROPERTY OF THEIR RES PECTIVE OWNERS. © 2022 NXP B.V.
J U N E 2 0 2 2
BL Edge Processing
NXP EDGEREADY
SMART TOUCHLESS HMI
SOLUTION OVERVIEW
1PUBLIC
i .MX RT117H MCU BASED SMART TOUCHLESS HMI SOLUTION
i.MX
RT117H
K32W41A
M/061
IW416
Machine Vision, Voice Control and Graphics Display UI all on one MCU Platform for Advanced and Innovative HMI
Image Sensor
RGB 720p
Image Sensor
IR VGA
Digital
Microphone
Digital
Microphone
Speaker
Flash
SDRAM
MikroE Click™
expansion header
Status Dates
Customer Sample Mar 2022
Market Launch Jun 2022
Solution Kit: SLN-TLHMI-IOT
RSL $299
Graphics Display User Interface
• 5.5-inch MIPI LCD display at 720*1280 resolution
• 2D graphics accelerator supported by LVGL
• GUI demo for Coffee Machine, Smart Home Panel and Elevator
Machine Vision
• 720p RGB and VGA IR Image Sensors
• Support user identification through face recognition
• Support gesture control through palm shape recognition
Voice Control
• 2 digital microphones on board and another 2 more through expansion
• VoiceSeeker integrated to support acoustic echo cancellation, noise reduction, beamforming and barge-in
• VIT for wake word and local voice command (English, Mandarian)
• Cyberon for wake word and local voice command with more languages support and separated charge through 3rd party
Wireless Connectivity
• Dual band Wi-Fi 4 + BLE combo: IW416
• Cloud based device management, user data management, OTA, etc.
Misc.
• 32MB flash, 32MB SDRAM, 2W louder speaker
• MikroE Click™ Expansion Header, PIR sensor
2PUBLIC
F U T U R E P O S S I B I L I T I E S
Smart Touchless HMI Machine Vision Remote Conference Payment
• Large screen display with 2D GUI
• User identification
with face recognition
• Touchless control with gesture
recognition (hands, head)
• Touchless control with far field
offline voice control
• Wireless connectivity
• 3D or 2D camera input
• Object detection and/or identification
• Human posture or
drowsiness detection
• Wired or wireless connectivity
• Speaker identification
with face recognition
• Speaker face detection and tracking
• Far field full-duplex communication
• Far field voice control
• Wired or wireless connectivity
• Face recognition with liveness
detection for payment
• Wireless connectivity
• EdgeLock security for payment
SMART TOUCHLESS HMI TARGET APPLICATIONS
5PUBLIC
i .MX RT117H MCU-BASED SMART TOUCHLESS HMI SOLUTION
• Development kit SLN-TLHMI-IOT – $299
• MCU
− i.MX RT117x (1 GHz Cortex-M7 + 400 MHz M4 Cores)
− 2 MB on chip SRAM
• Camera(s)
− RGB + IR (optional) CMOS Sensors
• Radio
− Dual band Wi-Fi 4 + BLE (NXP IW416)
• Display driven by LVGL graphics
− 5.5” LCD Touchscreen (720 x 1280)
• Audio hardware
− 2x DMICs + Speaker + Amplifier
• MikroE Click™ Expansion Header
− E.g. add NFC, more DMICs, IEEE 802.15.4 radio (Zigbee/Thread/Matter), etc.
6PUBLIC
FACE RECOGNIT ION TOUCHLESS HMI SOLUTION DEVELOPMENT KIT (SLN -TLHMI - IOT)
HARDWARE BLOCK DIAGRAM
32 MB QSPI
NOR FLASH
32 MB SDRAM
PC200
QSPI SEMC
Antenna
WHITE LEDPWM
LED DRIVER
LOW COST
LDO(s)
USB-C (5V)
CONNECTOR
Wi-Fi/BLEIW416
24 MHz
CRYSTAL
USER/RESET
SWITCH
MCU
i.MX RT117H
PDM
3.3 V BUCK
REG
EXT. 2xMIC
HEADER
SDIO
UART
GPIO
PIR
SENSOR
RGB
STATUS LED
GPIO
GPIO
GPIO
IR LEDPWM
LED DRIVER
IMAGE SENSOR
(HIGH RES. RGB)
DIGITAL
MICROPHONE
DIGITAL
MICROPHONE
PDM
MQSAUDIO OUT
AUDIO
AMPLIFIER
MIPI-DSIHIGH RESOLUTION
COLOR DISPLAY I2C / GPIO
IMAGE SENSOR
(RGB)
EXPANSION
HEADER
UART SPI I2C
ADC PWM GPIOs
IMAGE SENSOR
(IR)
FLEXIO
I2C / GPIO
MIPI-CSI
I2C / GPIO
DVP-CSI
I2C / GPIO
7PUBLIC
Fre
eR
TO
S
OTA FW Update
Boot Loader
Face Detection
NXP Edge Ready Machine Vision Library
NXP Facial Biometric Authentication & Gesture Recognition API
NXP MCU inference engine
Face Alignment
Face Quality 2D Liveness Algorithm
Face Recognition
Memory Management
Min
i C
V
Beamforming
Echo
Cancellation
and Barge-in
Full duplex
speaker phone
Audio Front End Runtime Library
Automatic Speech Recognition
Wake Words & Commands
ML Speech Engine Runtime Library
Gesture Recognition
Hardware Abstraction Layer
Algorithm Manager
HAL
Output Manager
HAL
FrameworkAlgorithm
ManagerOutput
Manager
Multicore
Manager HAL
Multicore
Manager
VIT (Voice
Intelligent
Technology)
or
Hardware Abstraction Layer
Display Manager
HALCamera Manager
HAL
Framework
Display
ManagerCamera
Manager
Multicore
Manager HAL
Multicore
Manager
Input
Manager
Output
Manager
Power Manager
Manager
Input Manager
HAL
Application Control
GUI Shell Wi-Fi/BLE Voice Input
Camera
Fle
xIO
MIP
I-C
SI2
CS
I
I2C
Display
LC
DIF
MIP
I-D
SI
GPU
PxP
GP
U2D
LED
PW
M
Wi-Fi
SD
IO
BLE
UA
RT
Dual Core
MU
SE
MA
4
SH
AR
ED
ME
MO
RY
Touch
I2C
Dual Core
MU
SE
MA
4
SH
AR
ED
ME
MO
RY
DMA
PD
M t
o
PC
M
Speaker
MQ
S
CM7 Driver Layer
CM4 Driver Layer
MSD FW Update
FLASH
QS
PI
File-system
Serial-manager
lwIP
mBedTLS
MQTT
LVGL
VGLite
H.264
Multi-Core
Manager
Opus decode
Middleware
Output Manager
HALPower Manager
HAL
SLN-TLHMI - IOT ( i .MX RT117H) EDGEREADY SOLUTION SOFTWARE BLOCK DIAGRAM
8PUBLIC
SLN-TLHMI - IOT ( i .MX RT117H) EDGEREADY SOLUTION SOFTWARE ARCHITECTURE
IR & RGB
Frames
CM7 CM4IPC with Shared Memory
Framework
MQS HAL
Vision Algo HAL
Voice Algo HAL
Peripherals
Voice Algo
Result
Notification
Message
Framework
Camera HAL
Display HAL
Graphics HAL
LVGL & VGLite HAL
Vision Algo
Result
Display
Camera
PxP
Touch
GPU2D
Peripherals
Mic
Speaker
• CM7 (Vision & Voice algorithm accelerator):
− Framework
− Vision algorithm with VGA input frames
− Voice algorithm (AFE + ASR) with mic input
− MQS audio playback
− IPC communication with shared memory
• CM4 (UI & System control unit):
− Framework
− CSI/MIPI Camera preview @VGA
− LVGL GUI @720p with VGLite 2D GPU acceleration
− Vision algorithm input frames color space conversion with PxP
− Touch panel input
− IPC communication with shared memory
9PUBLIC
SMART TOUCHLESS HMI SOLUTION USE CASE EXAMPLES FROM
• Face recognition for user
identification & customization
• Hand position and
gesture recognition
• GUI display, voice control and machine vision enabled with single MCU solution
• Combines face recognition & voice control from i.MX RT106F & i.MX RT106S
showroom.nxp.com
1 0PUBLIC
NXP EDGEREADY FACE RECOGNIT ION SOLUTIONS – ADVANTAGES SUMMARY
Low cost• MCU implementation reduces BOM cost as much as 50% vs. MPU + Linux® solutions
• Not an add-on solution - allows developer to replace host MCU with i.MX RT
− Available MHz & memory, and ability to trade off inference time vs. CPU load
Quick & easy
to use
• Turnkey solution – examples of concept to production in only four months
• Familiar RTOS based environment for IoT developers
• No face recognition, speech recognition, audio or ML/AI expertise necessary
• One stop shop, no third-party software
• MCU + RTOS solution minimizes boot time, enabling deep sleep for long battery life
• Plug and play out-of-box-experience
• Worldwide distribution network for sales and support
• Easy to use development tools (MCUXpresso IDE, SDK, configuration & provisioning)
Flexible
• Choose between low-cost CMOS sensors or high-performance 3D SLM cameras
• Common platform for smart lock, touchless HMI, and more to come…
• Offline capabilities lower latency & total cost of ownership, while preserving privacy
• Remote registration capability available on mobile and PC platforms