Securing and Enhancing The Resilience of Ai-based Embedded Control Modules for Autonomous Drone Squads in Hostile Environments

**Securing and Enhancing the Resilience of AI-Based Embedded Control Modules for Autonomous Drone Squads in Hostile Environments**:

- Réf **ABG-132334**
- Sujet de Thèse
- 03/06/2025
- Autre financement public
- Grenoble INP - LCIS
- Lieu de travail- Valence - Auvergne-Rhône-Alpes - France
- Intitulé du sujet- Securing and Enhancing the Resilience of AI-Based Embedded Control Modules for Autonomous Drone Squads in Hostile Environments
- Champs scientifiques- Informatique
- Electronique
- Mots clés- embedded AI, resilient control, fault tolerance, hardware security, fault injection, side-channel, autonomous drones, cyber-physical systems

**Description du sujet**:
ESSAIM-CPS Project — Context and Objectives

The ESSAIM-CPS project aims to improve the reliability, resilience, and security of autonomous drone squads operating in uncertain and hostile environments. Each drone incorporates AI-based control modules for critical tasks such as trajectory tracking, obstacle avoidance, formation maintaining, and energy management. These AI modules are vital for mission success, but introduce new vulnerabilities to the system:

- Environmental disturbances: unpredictable obstacles, weather conditions.
- Internal faults: faulty sensors, hardware failures, calculation errors.
- Physical or cyber-attacks aimed at AI modules (through fault injection or malicious data inputs) could change control decisions and affect squad performance.

The ESSAIM-CPS project’s goal is to:

- Design fault-tolerant and disturbance-resilient control architectures for embedded systems.
- Secure and harden AI-based control modules against attacks and perturbations to preserve decision integrity.
- Detect anomalies, attacks, and faults affecting AI control modules and limit their propagation across the system.
- Maintain squad operational capacity under degraded conditions, even when one or several AI modules are compromised.

PhD Objectives:

- Analyze hardware vulnerabilities of AI-based embedded control modules.
- Develop fault-tolerant and disturbance-resilient AI-based control techniques.
- Design embedded anomaly and attack detection mechanisms for AI control modules.
- Integrate these countermeasures into a robust, distributed control architecture capable of sustaining mission performance and safety even under attacks or AI module failures.
- Validate proposed methods through simulation and real drone deployment.

Keywords: embedded AI, resilient control, fault tolerance, hardware security, fault injection, side-channel, autonomous drones, cyber-physical systems.

**PhD Student Profile (any of the following)**:

- Master in Embedded Systems or Microelectronics
- Master in Computer Science or cybersecurity

**Required Skills**:

- Embedded Artificial Intelligence
- Embedded Systems Security
- Embedded Programming and Development
- Optional (a plus): Embedded Control and Command

**Prise de fonction**:

- 01/10/2025

**Nature du financement**:

- Autre financement public

**Précisions sur le financement**:
**Présentation établissement et labo d'accueil**:

- Grenoble INP - LCIS

The LCIS (Laboratory of Design and Integration of Systems) is a public research laboratory affiliated with Grenoble INP - UGA and located in Valence, France. The lab specializes in the design of secure, reliable, and intelligent embedded systems.

Its research activities focus on several key areas:
Embedded systems security (hardware and software, fault injection, side-channel attacks, countermeasures)

Cyber-Physical Systems (CPS)

Real-time and autonomous systems

LCIS actively collaborates with academic and industrial partners in the fields of aerospace, automotive, IoT, and defense. It is involved in national and European research projects and supports technology transfer to industry.

The lab hosts around 80 researchers, faculty members, PhD students, and engineers in a dynamic and collaborative environment.

**Site web**:
**Intitulé du doctorat**:

- Microélectronique et nanotechnologie

**Pays d'obtention du doctorat**:

- France

**Etablissement délivrant le doctorat**:

- UGA

**Ecole doctorale**:

- Electronique, électrotechnique, automatique, traitement du signal

PhD Student Profile (any of the following): - Master in Embedded Systems or Microelectronics - Master in Computer Science or cybersecurity Required Skills - Embedded Artificial Intelligence - Embedded Systems Security - Embedded Programming and Development - Optional (a plus): Embedded Control and Command- 06/06/2025