PD Screen AI

PD.Screen_AI family is an industrial-grade (-30…+80 °C) outdoor Smart Touch HMI platform built on a powerful AI MCU architecture alongside ESP32 MCUs—engineered to bring real-time edge AI intelligence into human-machine interaction (HMI).

This is not just an HMI.
It is a multi-modal AI interface system capable of hearing, seeing, understanding, and responding—directly on-device.

AI MCU Performance (Neural Networks & LLMs)

Neural Network Performance

• Up to ~600 GOPS (INT8 optimized inference)
• Dedicated NPU (Neural Processing Unit) for efficient AI workloads
• Optimized for CNNs, audio models, and lightweight transformers
• Real-time Vision (object detection, classification)
• Audio (keyword spotting, sound classification)
• Sensor data (anomaly detection)

Tiny LLM / Edge Language Capabilities

• Support for compact LLMs (TinyML / edge transformers)
• Typical deployment range:
  • 1M – 20M parameter models (fully on-device)
• Advantages for real use cases:
  • Voice command understanding
  • Intent classification
  • Context-aware UI responses
  • Ultra-low latency inference (milliseconds-level response)
  • Fully offline operation → no cloud, no data leakage

Efficiency Advantages

• 10–100× lower power consumption vs Linux-based AI platforms
• Deterministic execution (no OS scheduling jitter)
• Optimized memory footprint for embedded deployment

AI That Interacts With the Real World

PD.Screen_AI transforms traditional interfaces into intelligent assistants through:

Audio Intelligence

• Real-time audio recognition (voice commands, keywords, alarms)
• On-device speech processing without cloud dependency
• Audio generation / voice feedback for interactive systems
• Noise-resilient operation in industrial environments

Accessibility & Inclusive Design

• Voice-controlled interfaces for visually impaired users
• Audio guidance and feedback for hands-free operation
• Smart UI adaptation based on user interaction patterns

Vision & Object Recognition

• Object detection and classification (tools, packages, products)
• Scene understanding for automation workflows
• Real-time event detection and anomaly recognition

Gesture Recognition

• Touchless control using hand gestures
• Safe operation in sterile, industrial, or hazardous environments
• Reduced wear compared to physical interfaces

Predictive Maintenance Intelligence

• Continuous monitoring of sensor and system data
• Anomaly detection and failure prediction
• Audio + vibration pattern analysis for early fault detection
• Reduced downtime and service costs

Direct Vision Pipeline (MIPI CSI → AI)

PD.Screen_AI integrates a dedicated MIPI CSI camera interface, enabling a high-speed, low-latency video stream directly into the AI MCU.

• Native MIPI CSI camera support (no external processing required)
• Direct video pipeline into AI engine (no bottlenecks, no OS layers)
• Real-time vision inference on live video streams
• No need for external SoCs, GPUs, or Linux systems
• Ideal for object detection, gesture recognition, and scene analysis

This architecture enables true edge vision AI, where data is processed instantly and locally—ensuring privacy, reliability, and deterministic performance.

Why PD.Screen_AI Is Different

Traditional solutions (Raspberry Pi, generic MCUs, or Linux-based HMIs) struggle with:

• Latency and cloud dependency
• OS instability and corruption risks
• Complex AI deployment pipelines

PD.Screen_AI delivers deterministic, real-time AI at the edge.

Key Advantages:

• Instant boot (milliseconds)
• No heavy OS, no filesystem → maximum reliability
• On-device AI inference (audio, vision, gestures)
• Direct camera-to-AI pipeline via MIPI CSI
• Deterministic real-time response
• Energy-efficient edge processing
• Fastest time-to-market in its class
• Remote firmware updates (I2C / UART)

Hardware & Display

Supports 7” to 22” industrial TFT displays, combined with optimized firmware delivering:

• Smooth, real-time GUI
• Native touch + gesture interaction
• AI-enhanced UI responsiveness

ESP32 remains fully accessible (Arduino, PlatformIO, ESP-IDF), enabling:

• Rapid prototyping
• Wireless connectivity (WiFi / Bluetooth)
• Custom application development

Industrial Integration Ready

Designed as a central intelligent node, supporting:

• Microphone and speakers for audio AI
• MIPI CSI cameras for vision AI
• QR/barcode scanners (1D/2D)
• Motor drivers (automation systems)
• Thermal printers
• RS-485, CAN, UART, I2C
• Ethernet + PoE
• WiFi / Bluetooth
• GPIO expansion
• RTC with battery backup

From Interface to Intelligent System

PD.Screen_AI replaces traditional HMI stacks with:

• No driver development
• No complex AI infrastructure
• No cloud dependency required

Unified API-controlled AI + GUI platform
Real-time audio + vision fusion
Rapid transition from prototype to production

High-Impact Use Cases

Assistive & Accessibility Systems

• Voice-guided kiosks
• Interfaces for visually impaired users
• Smart assisted-living control panels

Industrial & Predictive Maintenance

• Machine health monitoring
• Acoustic + visual fault detection
• Smart industrial dashboards

Smart Infrastructure

• AI-powered lockers and parcel systems
• Contactless kiosks with gesture + voice control

Intelligent Automation

• Object-aware control panels
• Vision-guided workflows
• Human-machine collaboration interfaces

PD.Screen_AI is a next-generation edge AI HMI platform that combines audio intelligence, vision processing, and real-time control—with a direct camera-to-AI pipeline—to create systems that truly understand and interact with the physical world.