Skip to content

Commit 658afbb

Browse files
tokk-nvtokk-nv
authored
Merge pull request #332 from tokk-nv/feat/add-community-projects-2025
feat: Add 5 new community projects from 2025
2 parents 553f2c5 + 80a4ad6 commit 658afbb

File tree

6 files changed

+173
-11
lines changed

6 files changed

+173
-11
lines changed

‎src/content/projects/curious-frame-ai-tutor.md‎

Lines changed: 33 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,33 @@
1+
---
2+
title: "The Curious Frame: An Offline AI Tutor for Kids"
3+
description: "A screen-free educational device using Jetson Orin Nano with VLM and LLM to identify objects and explain them to kids in their language through voice."
4+
author: "Frederic Collonval"
5+
date: "2025-08-19"
6+
source: "Hackster"
7+
image: "https://hackster.imgix.net/uploads/attachments/1872832/_z3jxuwvChS.blob?auto=compress%2Cformat&w=900&h=675&fit=min"
8+
link: "https://www.hackster.io/frederic-collonval/the-curious-frame-an-offline-ai-tutor-for-kids-686677"
9+
featured: true
10+
tags: ["education", "kids", "vlm", "llm", "moondream", "gemma", "tts", "ollama", "orin-nano", "screen-free"]
11+
---
12+
13+
Frederic Collonval created a screen-free AI tutor for children that uses a handheld cardboard frame to identify everyday objects and explain them through voice in multiple languages, all running locally on Jetson Orin Nano.
14+
15+
The system uses moondream2 VLM for object detection, Gemma 3n LLM via Ollama for generating age-appropriate explanations, and Piper TTS for voice output. A clever flag-based interface allows children to switch between languages, turning it into both an educational tool and a language-learning device.
16+
17+
### Features
18+
19+
- Screen-free interaction for healthy child development
20+
- moondream2 vision-language model for object detection
21+
- Gemma 3n LLM for child-safe, age-appropriate explanations
22+
- Piper TTS for natural voice narration
23+
- Multilingual support with visual language switching
24+
- Fully offline operation on Jetson Orin Nano 8GB
25+
- Simple cardboard frame interface requiring no complex input
26+
- Automated setup script for easy deployment
27+
28+
### Resources
29+
30+
- [Hackster Project](https://www.hackster.io/frederic-collonval/the-curious-frame-an-offline-ai-tutor-for-kids-686677)
31+
- [GitHub Repository](https://github.com/webscit/curious-frame)
32+
33+

‎src/content/projects/gps-denied-drone-vslam.md‎

Lines changed: 33 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,33 @@
1+
---
2+
title: "GPS-Denied Drone with NVIDIA Jetson Orin Nano"
3+
description: "Autonomous drone using real-time VSLAM on Jetson Orin Nano with Isaac ROS to navigate GPS-denied environments via visual-inertial odometry and onboard mapping."
4+
author: "Andrew Bernas"
5+
date: "2025-06-22"
6+
source: "Hackster"
7+
image: "https://hackster.imgix.net/uploads/attachments/1860199/_3us6r87BE8.blob?auto=compress%2Cformat&w=900&h=675&fit=min"
8+
link: "https://www.hackster.io/bandofpv/gps-denied-drone-with-nvidia-jetson-orin-nano-9f3417"
9+
featured: true
10+
tags: ["vslam", "isaac-ros", "drone", "autonomous", "px4", "mavros", "robotics", "orin-nano"]
11+
---
12+
13+
Andrew Bernas developed an autonomous UAV platform powered by Visual Simultaneous Localization and Mapping (VSLAM) using Isaac ROS, integrated with MAVROS and PX4 flight stack. The system enables real-time localization and mapping for autonomous navigation in GPS-denied environments.
14+
15+
By fusing visual-inertial data from an Intel D435i camera and leveraging Jetson Orin Nano's edge computing capabilities, the drone can maintain situational awareness, dynamically map unknown environments, and execute autonomous flight patterns with high precision—all without GPS dependency.
16+
17+
### Features
18+
19+
- Isaac ROS VSLAM for real-time visual-inertial odometry
20+
- PX4 flight controller integration via MAVROS
21+
- Intel RealSense D435i stereo camera with IMU
22+
- Extended Kalman Filter (EKF) for sensor fusion
23+
- Autonomous flight pattern execution (square, circle, figure-8, spiral)
24+
- Real-time mapping and localization on edge device
25+
- Complete ROS2 integration for visualization and control
26+
- Fully autonomous operation in GPS-denied scenarios
27+
28+
### Resources
29+
30+
- [Hackster Project](https://www.hackster.io/bandofpv/gps-denied-drone-with-nvidia-jetson-orin-nano-9f3417)
31+
- [GitHub Repository](https://github.com/bandofpv/VSLAM-UAV)
32+
33+

‎src/content/projects/gpt-oss-storyteller.md‎

Lines changed: 32 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,32 @@
1+
---
2+
title: "GPT-OSS-20B Interactive Storyteller"
3+
description: "An interactive storytelling system for children using Jetson AGX Orin with LLM, image generation, TTS, and voice control for creating dynamic stories with visuals and sound."
4+
author: "Timothy Lovett"
5+
date: "2025-09-23"
6+
source: "Hackster"
7+
link: "https://www.hackster.io/timo614/gpt-oss-20b-storyteller-c91042"
8+
image: "https://hackster.imgix.net/uploads/attachments/1889223/_IGMA29ZuBT.blob?auto=compress%2Cformat&w=900&h=675&fit=min"
9+
featured: true
10+
tags: ["llm", "storytelling", "comfyui", "sdxl", "tts", "asr", "whisper", "ollama", "agx-orin"]
11+
---
12+
13+
Timothy Lovett created an interactive storytelling system for children that combines multiple AI models on Jetson AGX Orin to generate dynamic stories with accompanying visuals, voice narration, and adaptive sound effects.
14+
15+
The system uses voice input from children to shape story direction, generates illustrations on-the-fly with ComfyUI and SDXL, narrates with Piper TTS, and controls mood-based background music through a Seeed MIDI Synthesizer. All processing runs locally on the edge without cloud connectivity.
16+
17+
### Features
18+
19+
- GPT-OSS-20B LLM via Ollama for story generation
20+
- ComfyUI + SDXL with HyperSD LoRA for real-time image generation
21+
- Faster Whisper for voice-to-text transcription
22+
- Piper TTS for natural voice narration
23+
- Seeed MIDI Synthesizer for interactive control and mood-based audio
24+
- Projector display for immersive storytelling experience
25+
- Fully local processing on Jetson AGX Orin
26+
27+
### Resources
28+
29+
- [Hackster Project](https://www.hackster.io/timo614/gpt-oss-20b-storyteller-c91042)
30+
- [GitHub Repository](https://github.com/Timo614/gpt-oss-storyteller)
31+
32+

‎src/content/projects/n8n-rag-agent.md‎

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,31 @@
1+
---
2+
title: "Build Local AI RAG Agent with n8n on NVIDIA Jetson AGX Orin"
3+
description: "A fully local Retrieval-Augmented Generation (RAG) AI agent running on Jetson AGX Orin using n8n workflow automation, Qwen3-4B LLM, and Telegram integration."
4+
author: "Nurgaliyev Shakhizat"
5+
date: "2024-08-12"
6+
source: "Hackster"
7+
link: "https://www.hackster.io/shahizat/build-local-ai-rag-agent-with-n8n-on-nvidia-jetson-agx-orin-665914"
8+
image: "https://hackster.imgix.net/uploads/attachments/1872883/_kR3My3eE2X.blob?auto=compress%2Cformat&w=900&h=675&fit=min"
9+
featured: true
10+
tags: ["rag", "llm", "n8n", "workflow-automation", "qwen", "telegram", "agx-orin"]
11+
---
12+
13+
Nurgaliyev Shakhizat demonstrates how to build a fully local Retrieval-Augmented Generation (RAG) AI agent on the NVIDIA Jetson AGX Orin Developer Kit using n8n workflow automation platform.
14+
15+
The system integrates multiple components including Qwen3-4B LLM served via SGLang, Snowflake-Arctic-Embed2 for text embeddings through Ollama, Qdrant vector database for semantic search, and Telegram for real-time messaging interface. WireGuard VPN provides secure remote access without exposing services publicly.
16+
17+
### Features
18+
19+
- Fully offline RAG pipeline on edge device
20+
- n8n workflow automation for orchestration
21+
- Qwen3-4B LLM with SGLang for low-latency inference
22+
- Qdrant vector database for semantic retrieval
23+
- Telegram bot interface for natural interaction
24+
- Secure remote access via WireGuard VPN
25+
26+
### Resources
27+
28+
- [Hackster Project](https://www.hackster.io/shahizat/build-local-ai-rag-agent-with-n8n-on-nvidia-jetson-agx-orin-665914)
29+
- [NVIDIA Forums Discussion](https://forums.developer.nvidia.com/t/build-local-ai-rag-agent-with-n8n-on-nvidia-jetson-agx-orin/342095)
30+
31+

‎src/content/projects/reflective-diffusion-mirror.md‎

Lines changed: 33 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,33 @@
1+
---
2+
title: "Reflective Diffusion: AI Magic Mirror"
3+
description: "An interactive magic mirror using Jetson AGX Orin with SDXL diffusion to render artistic images from voice prompts and camera captures."
4+
author: "Timothy Lovett & Kerin Lovett"
5+
date: "2025-08-15"
6+
source: "Hackster"
7+
image: "https://hackster.imgix.net/uploads/attachments/1875147/_37gyfNJtG3.blob?auto=compress%2Cformat&w=900&h=675&fit=min"
8+
link: "https://www.hackster.io/irelas/reflective-diffusion-0824ae"
9+
featured: true
10+
tags: ["sdxl", "diffusion", "comfyui", "whisper", "image-generation", "mirror", "esp32", "agx-orin"]
11+
---
12+
13+
Timothy and Kerin Lovett created an interactive magic mirror that combines a wall-mounted display with real-time SDXL image generation. The system captures photos and voice prompts via ESP32S3, processes them through ComfyUI on Jetson AGX Orin, and renders artistic transformations on a mirror-enclosed screen.
14+
15+
The magic mirror uses mirror acrylic that appears as a normal mirror when off but displays generated images when active. Voice prompts are transcribed with Whisper, combined with camera captures, and processed through SDXL with ControlNet++ and Depth Anything V2 for coherent artistic renderings.
16+
17+
### Features
18+
19+
- Real-time SDXL image generation with ComfyUI
20+
- Voice-activated prompting using Whisper for transcription
21+
- ESP32S3 Sense for wireless camera and audio capture
22+
- ControlNet++ with Depth Anything V2 for depth-aware generation
23+
- HyperSD LoRA for fast 2-step diffusion
24+
- Mirror acrylic display in shadow box frame
25+
- WebSocket-based image streaming
26+
- Fully local processing on Jetson AGX Orin
27+
28+
### Resources
29+
30+
- [Hackster Project](https://www.hackster.io/irelas/reflective-diffusion-0824ae)
31+
- [GitHub Repository](https://github.com/Timo614/reflective-diffusion)
32+
33+

‎src/pages/community.astro‎

Lines changed: 11 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -25,7 +25,7 @@ const sourceColors: Record<string, { bg: string; text: string }> = {
2525
<div class="absolute -top-40 -right-40 w-96 h-96 bg-nvidia-green/20 rounded-full blur-3xl"></div>
2626
<div class="absolute top-20 -left-20 w-72 h-72 bg-purple-500/10 rounded-full blur-3xl"></div>
2727
</div>
28-
28+
2929
<div class="relative max-w-7xl mx-auto px-4 sm:px-6 lg:px-8 text-center">
3030
<h1 class="text-5xl md:text-6xl font-bold tracking-tight mb-6">
3131
<span class="bg-clip-text text-transparent bg-gradient-to-r from-nvidia-white to-nvidia-gray-300">
@@ -35,12 +35,12 @@ const sourceColors: Record<string, { bg: string; text: string }> = {
3535
<p class="text-xl text-nvidia-gray-300 max-w-2xl mx-auto leading-relaxed">
3636
Discover guides, tutorials, and articles contributed by the community showcasing generative AI on the Jetson platform.
3737
</p>
38-
38+
3939
<div class="mt-10 flex flex-wrap justify-center gap-4">
4040
<a href="https://discord.gg/nvidia-jetson" target="_blank" rel="noopener" class="px-6 py-3 bg-nvidia-green text-nvidia-black font-bold rounded-lg hover:bg-nvidia-green/90 transition-all">
4141
Join Discord
4242
</a>
43-
<a href="https://github.com/dusty-nv/jetson-containers" target="_blank" rel="noopener" class="px-6 py-3 bg-white/10 backdrop-blur-sm text-white font-semibold rounded-lg hover:bg-white/20 transition-all border border-white/10">
43+
<a href="https://github.com/nvidia-ai-iot/jetson-ai-lab" target="_blank" rel="noopener" class="px-6 py-3 bg-white/10 backdrop-blur-sm text-white font-semibold rounded-lg hover:bg-white/20 transition-all border border-white/10">
4444
Contribute on GitHub
4545
</a>
4646
</div>
@@ -58,16 +58,16 @@ const sourceColors: Record<string, { bg: string; text: string }> = {
5858
{/* Media */}
5959
<div class="relative aspect-video bg-nvidia-gray-100 overflow-hidden">
6060
{project.data.video ? (
61-
<iframe
61+
<iframe
6262
src={project.data.video}
6363
class="w-full h-full"
6464
frameborder="0"
6565
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture"
6666
allowfullscreen
6767
></iframe>
6868
) : project.data.image ? (
69-
<img
70-
src={project.data.image}
69+
<img
70+
src={project.data.image}
7171
alt={project.data.title}
7272
class="w-full h-full object-cover group-hover:scale-105 transition-transform duration-300"
7373
/>
@@ -78,13 +78,13 @@ const sourceColors: Record<string, { bg: string; text: string }> = {
7878
</svg>
7979
</div>
8080
)}
81-
81+
8282
{/* Source Badge */}
8383
<span class={`absolute top-3 left-3 px-2 py-1 text-xs font-bold rounded ${colors.bg} ${colors.text}`}>
8484
{project.data.source}
8585
</span>
8686
</div>
87-
87+
8888
{/* Content */}
8989
<div class="p-5">
9090
<div class="flex items-center justify-between mb-2">
@@ -93,15 +93,15 @@ const sourceColors: Record<string, { bg: string; text: string }> = {
9393
{new Date(project.data.date).toLocaleDateString('en-US', { month: 'short', day: 'numeric', year: 'numeric' })}
9494
</span>
9595
</div>
96-
96+
9797
<h3 class="text-lg font-bold text-nvidia-black mb-2 group-hover:text-nvidia-green transition-colors line-clamp-2">
9898
{project.data.title}
9999
</h3>
100-
100+
101101
<p class="text-nvidia-gray-600 text-sm line-clamp-3">
102102
{project.data.description}
103103
</p>
104-
104+
105105
{project.data.tags && (
106106
<div class="flex flex-wrap gap-1.5 mt-4">
107107
{project.data.tags.slice(0, 3).map((tag: string) => (

0 commit comments

Comments
 (0)