matlab-deploy-embedded-ai

Name: matlab-deploy-embedded-ai
Author: matlab

815stars105forks

How to install matlab-deploy-embedded-ai?

matlab/matlab-agentic-toolkit/matlab-deploy-embedded-ai

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-deploy-embedded-ai

Installs into the current project.

Summary

Deploy AI models to embedded hardware using MathWorks tools (MATLAB, Simulink, Embedded Coder). Covers two workflow patterns: (1) MathWorks-native or imported models rebuilt as dlnetwork for lean hardware, (2) direct C/C++ code generation from PyTorch and LiteRT models. Both patt

Files · 1

View on GitHub

SKILL.md

1	# Embedded AI for Engineered Systems
2
3	Deploy AI models to embedded hardware using MATLAB® and Simulink®. This skill is
4	written specifically for MATLAB R2026a and uses APIs, functions, and workflows
5	introduced in that release. It covers the complete lifecycle: model creation or
6	import, verification, compression, system-level simulation, and code generation
7	for resource-constrained targets.
8
9	Requires MATLAB R2026a or newer. Core toolboxes: Deep Learning Toolbox, Statistics
10	and Machine Learning Toolbox, MATLAB Coder, Embedded Coder, Simulink, and
11	Fixed-Point Designer. Workflow-specific support packages are checked during
12	Environment Discovery. The MATLAB and Simulink Agentic Toolkits must be available
13	so the agent can drive a live MATLAB and Simulink session through MCP tools.
14
15	## When to Use
16
17	- Deploying a trained neural network (MATLAB-native or imported) to embedded hardware
18	- Generating C or CUDA code from a deep learning model for ARM Cortex-M/A/R, x86, or GPU targets
19	- Importing PyTorch, ONNX, TensorFlow, or LiteRT models into MATLAB for embedded deployment
20	- Compressing AI models (quantization, pruning, projection) to fit resource-constrained hardware
21	- Integrating AI inference into a Simulink system model for closed-loop simulation before code generation
22	- Using `loadPyTorchExportedProgram`, `importNetworkFromPyTorch`, `importNetworkFromONNX`, `importNetworkFromTensorFlow`, `dlquantizer`, `exportNetworkToSimulink`, or Embedded Coder with AI models
23	- Choosing between MathWorks-native code generation and direct PyTorch/LiteRT code generation
24
25	## When NOT to Use
26
27	- Training a model purely for research with no deployment target — use Deep Learning Toolbox documentation directly
28	- Deploying to cloud/server endpoints (no embedded target) — use MATLAB Production Server or MATLAB Compiler SDK
29	- Working with classical ML models (decision trees, SVMs, ensembles) that aren't neural networks — use Statistics and Machine Learning Toolbox codegen workflows directly. Note: `fitcnet`/`fitrnet` neural network models ARE covered by this skill
30	- Generating code for non-AI Simulink models — use standard Embedded Coder workflows
31	- Converting between model formats without an embedded deployment goal (e.g., ONNX to MATLAB for desktop inference only)
32
33	## Workflow Pattern Selection
34
35	This skill uses two deployment patterns:
36
37	- Pattern 1 — MATLAB Network Codegen: Import or build a `dlnetwork`, optionally
38	compress it, then generate C/C++ via MATLAB Coder or export to Simulink.
39	See [`references/pattern1/workflow.md`](references/pattern1/workflow.md).
40	- Pattern 2 — PyTorch/LiteRT Direct Codegen: Load a PyTorch (.pt2) or LiteRT
41	(.tflite) model directly and generate C/C++ without converting to a dlnetwork.
42	See [`references/pattern2/workflow.md`](references/pattern2/workflow.md).
43
44	### Decision Tree
45
46	Primary discriminator for external models: deployment capabilities + hardware class.
47
48	```
49	Q1: Where does the AI model come from?
50	\|
51	+-- Trained in MATLAB, or requires training in MATLAB -------> Pattern 1
52	\|
53	+-- External framework (PyTorch, TF, ONNX, Keras) --> Q2
54	\|
55	Q2: Does the deployment need any of these?
56	\| - Quantization (INT8 via dlquantizer)
57	\| - Pruning or projection
58	\| - Weight inspection / modification
59	\| - exportNetworkToSimulink integration
60	\|
61	+-- YES --> Pattern 1 (import as dlnetwork)
62	\|
63	+-- NO ---> Q3
64	\|
65	Q3: What is the deployment target?
66	\|
67	+-- Cortex-M: Pattern 1
68	\| (compression and Simulink verification typically needed)
69	\|
70	+-- x86 / GPU:
71	\| +-- PyTorch (.pt2) or LiteRT (.tflite) --> Pattern 2
72	\| +-- ONNX, TF, Keras --> Pattern 1 (convert to ONNX recommended)
73	\|
74	+-- Cortex-A/R:
75	+-- Small model --> Pattern 1
76	\| (import as dlnetwork, then codegen)
77	+-- Large model:
78	+-- PyTorch (.pt2) or LiteRT (.tflite) --> P

People who liked this also viewed

6 picks

Type

matlab-deploy-embedded-ai

815stars105forks

View on GitHub

How to install matlab-deploy-embedded-ai?

matlab/matlab-agentic-toolkit/matlab-deploy-embedded-ai

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-deploy-embedded-ai

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Embedded AI for Engineered Systems
2
3	Deploy AI models to embedded hardware using MATLAB® and Simulink®. This skill is
4	written specifically for MATLAB R2026a and uses APIs, functions, and workflows
5	introduced in that release. It covers the complete lifecycle: model creation or
6	import, verification, compression, system-level simulation, and code generation
7	for resource-constrained targets.
8
9	Requires MATLAB R2026a or newer. Core toolboxes: Deep Learning Toolbox, Statistics
10	and Machine Learning Toolbox, MATLAB Coder, Embedded Coder, Simulink, and
11	Fixed-Point Designer. Workflow-specific support packages are checked during
12	Environment Discovery. The MATLAB and Simulink Agentic Toolkits must be available
13	so the agent can drive a live MATLAB and Simulink session through MCP tools.
14
15	## When to Use
16
17	- Deploying a trained neural network (MATLAB-native or imported) to embedded hardware
18	- Generating C or CUDA code from a deep learning model for ARM Cortex-M/A/R, x86, or GPU targets
19	- Importing PyTorch, ONNX, TensorFlow, or LiteRT models into MATLAB for embedded deployment
20	- Compressing AI models (quantization, pruning, projection) to fit resource-constrained hardware
21	- Integrating AI inference into a Simulink system model for closed-loop simulation before code generation
22	- Using `loadPyTorchExportedProgram`, `importNetworkFromPyTorch`, `importNetworkFromONNX`, `importNetworkFromTensorFlow`, `dlquantizer`, `exportNetworkToSimulink`, or Embedded Coder with AI models
23	- Choosing between MathWorks-native code generation and direct PyTorch/LiteRT code generation
24
25	## When NOT to Use
26
27	- Training a model purely for research with no deployment target — use Deep Learning Toolbox documentation directly
28	- Deploying to cloud/server endpoints (no embedded target) — use MATLAB Production Server or MATLAB Compiler SDK
29	- Working with classical ML models (decision trees, SVMs, ensembles) that aren't neural networks — use Statistics and Machine Learning Toolbox codegen workflows directly. Note: `fitcnet`/`fitrnet` neural network models ARE covered by this skill
30	- Generating code for non-AI Simulink models — use standard Embedded Coder workflows
31	- Converting between model formats without an embedded deployment goal (e.g., ONNX to MATLAB for desktop inference only)
32
33	## Workflow Pattern Selection
34
35	This skill uses two deployment patterns:
36
37	- Pattern 1 — MATLAB Network Codegen: Import or build a `dlnetwork`, optionally
38	compress it, then generate C/C++ via MATLAB Coder or export to Simulink.
39	See [`references/pattern1/workflow.md`](references/pattern1/workflow.md).
40	- Pattern 2 — PyTorch/LiteRT Direct Codegen: Load a PyTorch (.pt2) or LiteRT
41	(.tflite) model directly and generate C/C++ without converting to a dlnetwork.
42	See [`references/pattern2/workflow.md`](references/pattern2/workflow.md).
43
44	### Decision Tree
45
46	Primary discriminator for external models: deployment capabilities + hardware class.
47
48	```
49	Q1: Where does the AI model come from?
50	\|
51	+-- Trained in MATLAB, or requires training in MATLAB -------> Pattern 1
52	\|
53	+-- External framework (PyTorch, TF, ONNX, Keras) --> Q2
54	\|
55	Q2: Does the deployment need any of these?
56	\| - Quantization (INT8 via dlquantizer)
57	\| - Pruning or projection
58	\| - Weight inspection / modification
59	\| - exportNetworkToSimulink integration
60	\|
61	+-- YES --> Pattern 1 (import as dlnetwork)
62	\|
63	+-- NO ---> Q3
64	\|
65	Q3: What is the deployment target?
66	\|
67	+-- Cortex-M: Pattern 1
68	\| (compression and Simulink verification typically needed)
69	\|
70	+-- x86 / GPU:
71	\| +-- PyTorch (.pt2) or LiteRT (.tflite) --> Pattern 2
72	\| +-- ONNX, TF, Keras --> Pattern 1 (convert to ONNX recommended)
73	\|
74	+-- Cortex-A/R:
75	+-- Small model --> Pattern 1
76	\| (import as dlnetwork, then codegen)
77	+-- Large model:
78	+-- PyTorch (.pt2) or LiteRT (.tflite) --> P

People who liked this also viewed

6 picks

Type