darklinux commited on
Commit
e230eb3
·
verified ·
1 Parent(s): ef1bb40

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +78 -30
README.md CHANGED
@@ -8,48 +8,96 @@ colorFrom: gray
8
  colorTo: blue
9
  sdk: static
10
  ---
 
 
 
 
 
 
 
 
 
11
  ---
12
 
13
- ---
14
- **IA_AUT_ECO: The Augmented Autonomy Architecture (Open Source Deep Tech)**
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
 
16
- **IA_AUT_ECO** is a Deep Tech startup based near Paris, France, dedicated to redefining the use of Artificial Intelligence and Robotics. Our mission is to build robust, open-source solutions for the daily lives and autonomy of disabled and invalid persons.
 
 
 
17
 
18
- We are building systems that guarantee digital sovereignty, social inclusion, and uncompromising security.
19
 
20
- **Our Commitment: Open Source for All**
21
- We are firm believers that the most critical assistance technologies (AI, Robotics) must be auditable and enduring. Our architecture is designed on an ethical Open Core model: the technical foundations will eventually be made public to guarantee full trust and accountability.
22
 
23
- **The Core is Under Development**: Our **open-source systems** are currently in the initial design and hardening phase. We plan to open up the foundational repositories for community auditing and contributions at a later date.
24
- Uncompromising Security: A Sovereign Digital Shield
25
- Trust is our currency. Our **cybersecurity solutions are integrated by design**, exceeding regulatory requirements.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
26
 
27
- **The Winterhold Foundational Architecture**
28
- **The Winterhold ecosystem** is our comprehensive offering, built upon several critical components, all currently under active development:
29
 
30
- * **Hearthfire** : The Hardened Security OS. A core operating system (forked from *FreeBSD*) for local deployment and robotics, emphasizing Security by Design, Zéro Trust (ZTA), and PQC Readiness.
31
- * **MASAQ** : The AI Orchestrator. Manages the secure and auditable execution of Deep Learning models. Focuses on SLM optimization and Knowledge Graph management.
32
- * **Ysgrammor** : The Ethical Audit System. Provides complete observability (OpenTelemetry) for all system actions and AI decisions, delivering Proof of Intent and ensuring NIS2 compliance.
33
- * **Skald : The Cognitive Agent** : User-facing AI for simplification and cognitive augmentation (GNN, Semantic Synthesis).
34
- * **DAWNSTAR** : The Safe Mobility Agent. A ROS 2 module for the robotic wheelchair base, ensuring **NIS2/ISO/SBOM compliant** control and navigation.
35
- * **ESBERN** : The Augmented Robotic Agent (Companion). The robotic assistant that handles fine manipulation and interaction, working in synergy with DAWNSTAR on the NVIDIA Jetson platform.
36
- * **DRAGONBORN** : is the demonstration application for the **SKALD Cognitive Agent*, designed to showcase the transformative power of Deep Learning in providing real-time sensory and cognitive assistance.
37
- Deployed across mobile platforms using TensorFlow Lite (TFLite), **DRAGONBORN* transcends simple object recognition. It translates the complexity of the visual world into simple, actionable semantic instructions and synthesized audio.
38
 
39
- This application is our "proof of concept" to the world: a powerful, accessible, and ethical use of AI that requires minimal technical overhead.
 
40
 
41
- **Guaranteed Compliance and Auditability**
42
- Our projects integrate European and international standards from the very beginning: the AI Act, NIS2 Directive, and SBOM (Software Bill of Materials) are central to our engineering.
43
 
44
- **The Deep Tech Vision: AI and Robotics for Daily Life**
45
- Our approach goes beyond simple interfaces:
 
 
46
 
47
- Cognitive Augmentation: **Skald** utilizes Deep Tech to transform complex information into personalized semantic synthesis, making critical information instantly accessible.
48
- Robotic Autonomy: The **DAWNSTAR** and **ESBERN** system creates a cohesive, sovereign platform, guaranteeing the physical and digital security of users relying on advanced robotic assistance.
49
- Status and Next Steps
50
- This repository is currently utilized for internal engineering and initial development of the Winterhold architecture.
51
 
52
- We will release the core code (including Hearthfire, MASAQ, Ysgrammor, DAWNSTAR, and ESBERN) as they reach an auditable maturity level.
53
- Stay Updated: Follow this repository for official announcements regarding the opening of our projects to community audit.
54
 
 
55
  ---
 
8
  colorTo: blue
9
  sdk: static
10
  ---
11
+
12
+ # IA_AUT_ECO: The Fabless Architecture for Augmented Autonomy
13
+
14
+ **"Let's be Useful."**
15
+
16
+ IA_AUT_ECO is a French Deep Tech company dedicated to redefining assistive technology. We operate on a **Fabless model**: we design the sovereign architecture, the secure software, and the hardware standards, while certifying partners for manufacturing.
17
+
18
+ Our mission is to build the **Winterhold Ecosystem**: a secure, open-source, and ethical foundation for the autonomy of disabled and invalid persons. We provide the "Brain" and the "Nervous System" for the next generation of assistive robotics.
19
+
20
  ---
21
 
22
+ ## The Winterhold Framework: Sovereignty & Security
23
+
24
+ Winterhold is not just software; it is a comprehensive architecture designed for **Zero Trust (ZTA)** environments, **NIS2** compliance, and **Post-Quantum Cryptography (PQC)** readiness.
25
+
26
+ ### 1. Hearthfire (The Hardened OS)
27
+ The foundation of our security. A minimalist, hardened **FreeBSD Fork** compatible with **OCI (Open Container Initiative)** standards.
28
+ * **Role:** Provides a secure, deterministic environment for embedded AI and real-time control.
29
+ * **Security:** Implements kernel-level PQC encryption and strict process isolation.
30
+
31
+ ### 2. MASAQ & Skald (The AI Core)
32
+ * **MASAQ:** The secure orchestration engine that manages Deep Learning models on Edge devices (NVIDIA Jetson).
33
+ * **Skald:** The Cognitive Agent. It uses Small Language Models (SLMs) and Graph Neural Networks (GNNs) to transform complex data into simple, actionable intentions.
34
+
35
+ ### 3. Ysgrammor (The Ethical Audit)
36
+ A real-time observability system based on OpenTelemetry. It records the "Proof of Intent" for every AI decision, ensuring full auditability for medical and legal compliance.
37
+
38
+ ### 4. Delphine (The Sovereign Data Core)
39
+ An on-premise, ethical CRM and polyglot database designed to manage critical social and medical data without relying on public clouds.
40
+
41
+ ---
42
+
43
+ ## The Robotic Agents: Dawnstar & Esbern
44
+
45
+ We provide the certified designs and software stacks for two complementary hardware agents.
46
 
47
+ ### DAWNSTAR: The Safe Mobility Agent
48
+ A set of **ROS 2** packages and hardware standards for robotic wheelchairs.
49
+ * **Focus:** Critical safety, slope calculation, speed limiting, and semantic navigation.
50
+ * **Tech:** Uses PyTorch for vision and scikit-learn for anomaly detection.
51
 
52
+ ### ESBERN: The Robotic Companion
53
 
 
 
54
 
55
+ [Image of robotic arm on wheelchair]
56
+
57
+ A mobile manipulator agent (robotic arm) that works in synergy with Dawnstar.
58
+ * **Focus:** Fine manipulation (grasping, door opening) and environment interaction.
59
+ * **Hardware:** Powered by NVIDIA Jetson for heavy edge computing.
60
+
61
+ ---
62
+
63
+ ## Companion Software Suite: Holistic Cognitive Assurance
64
+
65
+ Beyond simple navigation, the Esbern/Dawnstar system runs a suite of "Holistic Modules" designed to act as a cognitive prosthesis for conditions like Alzheimer's, Autism, or heavy disability.
66
+
67
+ * **MNEMOSYNE (Memory):** A local RAG system that records and indexes daily life to answer queries like "Where did I put my glasses?" or "Who visited yesterday?".
68
+ * **VESTA (Routine):** Proactive ADL (Activities of Daily Living) monitoring. Detects degradation in habits (hydration, sleep) and alerts before a crisis occurs.
69
+ * **KAIROS (Reality):** Manages confusion and "sundowning" phases using validation therapy techniques and environmental adaptation (lighting/music).
70
+ * **HERMES (Social):** A real-time social decoder for aphasia and autism, whispering context about interlocutors and predicting sentence completion.
71
+ * **HYGIE (Pharma):** High-precision visual identification of medication intake and passive biometric monitoring to ensure compliance and safety.
72
+
73
+ ---
74
+
75
+ ## Dragonborn: Universal Access (Mobile Demo)
76
+
77
+ **Dragonborn** is our mass-market demonstration of the Skald Cognitive Agent.
78
+ * **Platform:** Android (Kotlin) / iOS.
79
+ * **Tech:** Runs optimized **TFLite** models via a **Go** backend.
80
+ * **Function:** Translates the visual world into simple semantic instructions and audio, proving that powerful AI can be accessible on standard smartphones without the full Winterhold stack.
81
+
82
+ ---
83
 
84
+ ## The Fabless Business Model
 
85
 
86
+ IA_AUT_ECO focuses on high-value engineering and certification.
 
 
 
 
 
 
 
87
 
88
+ 1. **Open Design:** We release 3D blueprints (STEP/CAD), electrical schematics, and core software components to the community.
89
+ 2. **Winterhold Certification:** We license the **Winterhold** brand and commercial software support to manufacturing partners who pass our rigorous **DAWNSTAR-QC** testing suite and adhere to our component standards (NVIDIA Jetson, specific IMUs, ISO-compliant materials).
90
 
91
+ ## Status
 
92
 
93
+ **Current Phase:** Internal Engineering & Architecture Hardening.
94
+ * Validating Hearthfire kernel modifications.
95
+ * Training Skald/Mnemosyne models on NVIDIA Jetson.
96
+ * Developing the Dragonborn mobile prototype.
97
 
98
+ ## Contact & Licensing
 
 
 
99
 
100
+ For partnerships, Fabless certification inquiries, or institutional licensing of the Winterhold framework (Municipalities, NGOs), please contact the IA_AUT_ECO team.
 
101
 
102
+ **Secure. Sovereign. Useful.**
103
  ---