The projects below were supervised in my role as an academic research lead and form the conceptual and technical groundwork for the systems I am now building in an industrial and startup context.
Achref Doula
PhD (2020–2025)
On Uncertainty-Aware Perception, Prediction & Planning
This PhD focused on uncertainty-aware learning and neuro-symbolic modeling for perception and forecasting in autonomous systems. The work combines structured uncertainty representations with interpretable models for motion reasoning.
Selected Systems & Code
-
CLEAR-Command
A framework for uncertainty-aware command and control in autonomous systems.
🔗 https://gitlab.com/achref.d/clear-command -
KRPS
A structured planning and reasoning system integrating uncertainty-aware perception with symbolic decision-making.
🔗 https://gitlab.com/achref.d/krps -
NeSyMoF – Neuro-Symbolic Motion Forecasting
A neuro-symbolic approach to motion forecasting that combines learned representations with explicit symbolic structure.
🔗 https://gitlab.com/achref.d/nesymof
My role: Scientific supervision, research direction, model design, and publication strategy.
Thomas Kreutz
PhD (2022–2025)
LiDAR- and IMU-Based Human-Centric Scene Understanding
This PhD investigated human-centric scene understanding using LiDAR-first perception, multimodal embeddings, and spatio-temporal modeling, with applications to activity recognition and crowd simulation.
Selected Systems & Code
-
UMOSMOTS
Unsupervised 4D LiDAR moving object segmentation in stationary settings using multivariate occupancy time series.
🔗 https://github.com/thkreutz/umosmots -
Crowd Orchestration Simulator
A simulation framework for modeling and orchestrating crowd dynamics with spatio-temporal spawn processes.
🔗 https://github.com/thkreutz/crowdorchestrationsim -
DeSPITE
Contrastive deep skeleton–pointcloud–IMU–text embeddings for advanced human activity understanding.
🔗 https://github.com/thkreutz/despite
My role: Scientific supervision, conceptual framing, representation design, and coordination with publication and collaboration partners.