ISL is offering a PhD Position
Event Based Sensing and Insect Vision Inspired Systems for Computationally Frugal Drone Swarm Control
Objective of the PhD Thesis
The thesis aims to imagine, design, implement and validate a bi modal (event + intensity) perception-control module that mimics the parallel pathways of insect compound eyes, and to demonstrate that this system can run on embedded aerial platforms with low-power dedicated processor while delivering end to end low latency behavior in relevant defense related scenarios. The ultimate goal is to prove that a single drone equipped with this module can reliably perform high speed obstacle avoidance, fast target interception and, once validated, that the same approach can be scaled to control a drone swarm with decentralized event driven coordination.
Description of the Project
The integration of event-based sensing and insect vision inspired systems represents a paradigm shift in autonomous drone control, offering unprecedented capabilities for low-level, fast, and computationally frugal command and control in complex environments. This emerging field addresses critical challenges in real-time obstacle avoidance, strike detection, and swarm coordination through biomimetic approaches that leverage the efficiency of biological visual systems.
ISL is studying potential applications of event-based sensors for more than 6 years based on PhD works, NATO research group contributions and experimentations. Promising results were obtained in optical flow computation, semi-active laser terminal guidance and fast-moving object detection and tracking. This new PhD project will pursue the previous PhD, studying drone interception application, by adapting the bi-modal model approach to a more insect inspired sensing and processing architecture. The main idea is to reduce further the computational load whereas accelerating the whole processing chain by applying state-of-the art knowledge of arthropod vision.
Event-based sensing and insect vision inspired systems represent a transformative approach to autonomous drone control, addressing critical challenges in real-time obstacle avoidance and swarm coordination through biomimetic solutions. Unlike traditional frame-based cameras that capture complete images at fixed intervals, event-based sensors such as dynamic vision sensors (DVS) only detect changes in light intensity as they occur, generating sparse, asynchronous data streams with microsecond precision. This approach dramatically reduces computational load by factors of 100-1000 while achieving superior temporal resolution exceeding 10 kHz equivalent frame rates and power consumption as low as 2 mW. Insect compound eyes provide the biological inspiration for these systems, offering wide-field motion detection with minimal computational resources through parallel processing architectures. Recent biomimetic implementations achieve panoramic fields of view reaching 180 degrees while maintaining compact form factors, enabling simultaneous wide-angle imaging and depth perception. The integration of compound eye principles with neuromorphic processing architectures implements spiking neural networks that mirror biological neural computation, achieving energy consumption reductions compared to conventional CMOS-based imaging. These systems could excel in complex environment navigation, successfully detecting fast-moving objects and implementing collision avoidance at high-speeds using only event-based sensing. In swarm applications, the decentralized nature aligns with swarm intelligence principles, enabling semi-independent operation while maintaining collective behavior through simple interaction rules. Based on the literature review, this PhD thesis will study the scalability of this single agent behavior to try demonstrating its relevancy in coordinating drone swarms in challenging defense scenarios while operating at real-time.
The outcome of this PhD project will be a simulated and experimentally verified, insect vision inspired sensing and processing pipeline that delivers at low power consumption, high speed analysis and maneuverability for scalable swarm intelligence, key capabilities for both civilian autonomous navigation and Defense oriented rapid response missions.
What we offer
- 🗓 Contract duration: 3 years
- 🌐 Working language: English (French and/or German appreciated, language courses and casual multi-lingual coffee meetings available)
- A PhD project within a dynamic and international collaborative environment
- Access to state-of-the-art experimental facilities
- Competitive salary
Localization
The PhD work will be performed mainly at ISL (Saint-Louis, France), with occasional visits to the partner laboratories.
Supervising team
Nicolas HUEBER – Intelligent Guidance and Control
5 rue du Général Cassagnou – 68301 Saint-Louis
nicolas.hueber@isl.eu