Leszek Siwik, Ph.D, DS.C

Status: Under Realization | Open for PhD Candidates

This project focuses on the design and analysis of neuromorphic systems based on custom spiking neuron models, with the goal of bridging software simulations and hardware implementations. It investigates how biologically-inspired neuron dynamics can be modeled, optimized, and later realized in analog or mixed-signal hardware.

Research areas include development of custom spiking neuron models (e.g., leaky integrate-and-fire variants), integration with neuromorphic frameworks such as Tonic and snnTorch, replacement and evaluation of standard neuron models for selected tasks, and extension towards network-level modeling of simple nervous systems. The project also explores hardware-oriented implementations of neuron models and their impact on efficiency, scalability, and biological plausibility.

Status: Under Realization | Open for PhD Candidates

This project focuses on developing lightweight AI-based cybersecurity mechanisms for resource-constrained distributed autonomous systems. It investigates how intelligent security models can operate efficiently under strict limitations in computation, energy consumption, and communication while maintaining reliable threat detection capabilities.

Research areas include ultra-lightweight neural architectures for embedded cybersecurity, distributed and collaborative anomaly detection, energy-aware AI inference strategies, and adversarial robustness of lightweight models. The objective is to design scalable and efficient security solutions that enable autonomous nodes to perform real-time threat detection while preserving system performance and resource efficiency.

Status: Under Realization | Open for PhD Candidates

This project focuses on verifying the integrity and correct runtime behaviour of AI models It investigates how deviations from expected model behaviour can be detected and how potential compromises can be identified during system operation.

Research areas include behavioural fingerprinting of AI models, runtime anomaly detection, peer-based and collaborative verification mechanisms, consensus-based integrity validation in multi-agent systems, and distributed inference of model compromise. The objective is to develop robust and scalable methods that enable autonomous systems to continuously verify the trustworthiness of AI components under realistic operational constraints.

Status: Under Realization | Open for PhD Candidates

This project focuses on distributed detection of cyber threats in dynamic networks where individual nodes operate under partial, local, and potentially delayed observations. It investigates how system components can collaboratively infer the global security state of the network without relying on centralized monitoring.

Research areas include local anomaly detection, collaborative information exchange between nodes, distributed inference of global threat patterns, robustness against compromised or misleading nodes, and trade-offs between detection accuracy, communication overhead, and scalability. The objective is to develop resilient and scalable mechanisms that enable distributed systems to reliably detect and reconstruct coordinated cyber threats using limited, decentralized knowledge.

Status: Under Realization | Open for PhD Candidates

This project focuses on autonomous coordination of cyber-attacks and defence mechanisms in distributed systems without centralized control. Research areas include multi-agent adversarial modeling, coordinated attack planning and propagation, distributed anomaly detection, and adaptive, decentralized defence strategies. The objective is to improve resilience of large-scale infrastructures by enabling systems to autonomously detect, respond to, and mitigate coordinated cyber threats.

Status: Under Realization | Open for PhD Candidates

This project focuses on developing embodied and cognitive AI agents capable of understanding context, intent, and mission goals in cooperative astronaut–robot workflows. Research areas include adaptive mission execution, multimodal comprehension, avatar-based interaction behaviors, and socially-aware collaboration patterns. The objective is to enable humanoid and non-humanoid robotic systems to function as intuitive and reliable teammates during long-duration missions.

Status: Under Realization | Open for PhD Candidates

This project investigates autonomous coordination among multiple embodied robots deployed in orbital and planetary environments. We design decentralized communication frameworks, consensus mechanisms, and optimization strategies for collective behaviors such as inspection, logistics, maintenance, and repair. The project also examines scalable collaboration models for autonomous robot teams and hybrid human–robot groups.

Status: Under Realization

This project develops systems for real-time monitoring of astronauts’ physiological, behavioral, and emotional states. We integrate multimodal sensor data with AI-driven analytics to assess fatigue, cognitive load, stress, and psychological resilience. The research also includes predictive modeling and conversational agents capable of providing personalized emotional support in isolated, long-duration missions.

Status: Under Realization

This project explores emotionally intelligent interactions between humans and embodied AI systems. We develop expressive avatars, personalized communication styles, and social-memory mechanisms that support long-term relational continuity. The goal is to enhance psychological comfort, mission cohesion, and emotional resilience during deep-space missions.

Status: Under Realization

This project focuses on intuitive, multimodal interfaces for human–robot collaboration. Research topics include gesture, gaze, spatial cues, tactile communication, non-verbal signaling, shared situational awareness, and robot–robot coordination protocols. The aim is to create communication frameworks where robots can naturally convey intent, uncertainty, and state information in dynamic environments.

Status: Under Realization

The project is focused on robust, low-latency and ultra-low-power AI systems — including AI-on-chip and neuromorphic approaches for lightweight onboard reasoning under severe power, bandwidth, and computational constraints allowing spacecraft, orbital platforms, and distributed space infrastructures to reason about threats, operate safely under uncertainty, and maintain mission continuity even when isolated from Earth or operating under a full communication blackout.

Status: Under Realization

This project addresses autonomous technical, logistical, and safety-critical operations under hazardous conditions such as depressurization, radiation, contamination, or mechanical instability. We develop AI systems capable of independent task execution, anomaly detection, contingency planning, and real-time decision-making with minimal ground support. The aim is to ensure operational stability and mission continuity in extreme and unpredictable environments.