Part Number:
Quantity:
Price per Unit (in USD):
Total Amt:
Log in to myMicrochip to access tools and benefits. Sign up in just one minute.
Maximize Your Experience: Reap the Personalized Advantages by Completing Your Profile to Its Fullest. Update Here
Stay in the loop with the latest from Microchip. Update your profile while you are at it. Update Here
Complete your profile to access more resources. Update Here
true
Edge AI/ML Inferencing on dsPIC33 DSCs
Build Smart Solutions With Our Affordable and Premium dsPIC® DSCs
Why Choose dsPIC DSCs for Edge Machine Learning?
dsPIC DSCs bring intelligence to the edge by combining real-time control with Digital Signal Processor (DSP) performance to accelerate ML inference in embedded systems at affordable cost.
Key ML-Enabling Features
Fast ML Execution
High-Speed DSP Engine
- Offers single-cycle, multi-operation Multiply-Accumulate (MAC) instructions
- Accelerates ML math and matrix operations
Parallel Data Access
Dual, Independent Data Memory Banks
- Enables concurrent reads
- Reduces latency in ML data processing
ML Acceleration
Optimized DSP Libraries
- Includes ready-to-use DSP routines on MPLAB® XC Compiler
- Speeds up interfacing and minimizes development time
Precision at Speed
Hardware Floating-Point Unit (FPU) in dsPIC33A
- Boosts complex ML math performance
- Delivers higher accuracy for demanding models
Time-Critical Performance
Multi-Context ISR Registers
- Facilitates fast interrupt servicing
- Accelerates time-critical ML tasks like anomaly detection
Minimum Latency
Tightly Coupled Peripherals
- Reduces letency between sensing, processing and actuation
- Offers real-time control and fast feedback loops
Target Applications
Featured Reference Designs
Gesture Recognition
This demonstration highlights ML-based gesture recognition powered by a dsPIC33CK Digital Signal Controller (DSC).
Using data from an onboard Inertial Measurement Unit (IMU) sensor, the application captures and classifies user gestures through a trained Machine Learning (ML) model that is optimized for the dsPIC33CK platform. This demo showcases the entire development journey—from motion sampling to model training and real-time inference—all running efficiently at the edge.
Predictive Maintenance
This reference design showcases how MPLAB® Machine Learning (ML) Development Suite and the dsPIC® DSC-based Low-Voltage Motor Control (LVMC) board can enable real-time predictive maintenance at the edge.
By monitoring the motor’s Iq current, a trained classification model identifies whether the motor is operating normally or exhibiting anomalies such as an unbalanced load or a broken bearing—all without any sensors and any need for cloud processing.
Tools and Software
- Demo & Evaluation Boards
- Development Environment
- Software Solutions
The Curiosity Platform Development Board (EV74H48A) is a full-featured development and demonstration platform enabling customers to explore the capabilities of the dsPIC33A Digital Signal Controller (DSC) family and also the PIC32A MCU family. The board operation requires the insertion of a 120-pin General Purpose Dual In-Line Module (DIM) containing a dsPIC33A DSC device or a PIC32A MCU device.
This board simplifies the evaluation of capacitive touch sensing using the dsPIC33C DSCs with core-independent touch capability or the device’s usage in automotive and industrial applications requiring CAN-FD, LIN or SENT communication. This board has on-board touch buttons, CAN/CAN FD, LIN and SENT transceivers, along with XPLAINED PRO and mikroBUS™ connectors to interface touch and other extension boards
The dsPIC33CH Curiosity Development Board (DM330028-2) is intended as a cost effective development and demonstration platform for the entire dsPIC33CH family of dual core high performance digital signal controllers.
MPLAB® Machine Learning Development Suite Model Builder is a plug-in for MPLAB X IDE that builds optimized Artificial Intelligence (AI) sensor recognition code for dsPIC33 DSCs. The plug-in fully automates each step of the AI and Machine Learning (ML) workflows using a growing library of advanced ML and AI algorithms that learn from raw data in the development phase.
MPLAB X IDE is a highly configurable software platform that provides powerful, free tools for developing, debugging and qualifying embedded designs that use microcontrollers and digital signal controllers.
MPLAB® Data Visualizer
MPLAB Data Visualizer is a program used to process and visualize data from a running embedded target. The program can be accessed from within MPLAB X IDE or as a stand-alone program. With MPLAB Data Visualizer, you can see how key data points in your application change during runtime and record them, such as visualizing values captured by a sensor on your development board.
DSP Libraries for Machine Learning
The DSP library for dsPIC33A DSCs is a set of ready-to-use DSP functions that are optimized for lower latency operation to enable the development of edge AI/ML applications. These DSP functions leverage the dsPIC33A DSC architecture to improve efficiency and precision.
The DSP library for dsPIC33A devices comes pre-installed with the MPLAB XC-DSC compiler, located in its install directory.
Shining a Light on Artificial Intelligence and Machine Learning
This class will take a journey through artificial intelligence and machine learning and how you can use it in a simple embedded application.
