Robin
De Lissnyder

Online chess forces players to stare at computer screens. Digitizing a physical board without bulky overhead cameras requires solving magnetic sensor baseline drift, tracking rapid piece movements during captures, and syncing clocks with minimal latency.

A dual-processor setup: an ESP32 microcontroller handles deterministic microsecond scanning across 64 Hall sensors and drives 152 LEDs, while a Raspberry Pi runs game state rules, Lichess connectivity, and Stockfish AI.

Physical piece movements also double as gestures to review games, restart matches, or train endgames directly on the wood.

Active exoskeletons need to estimate lumbar joint torque in real time during repetitive lifting (8–22 kg) to prevent Work-Related Musculoskeletal Disorders ($L5/S1$ spinal compression). Traditional biomechanical software requires complex manual conversions between sensor files and physics solvers.

Built an automated pipeline in Python using the OpenSim API to convert raw Xsens IMU motion capture and Cometa wireless EMG data into standardized OpenSim motion files, calculating ground-truth joint torques via Inverse Dynamics.

Integrated a Dockerized web studio where researchers can select neural network architectures (CNN, LSTM, GRU), run CUDA-accelerated TensorFlow training, and evaluate live predictions.

Automating parcel palletizing is difficult because incoming conveyor packages vary in size, weight, and sequence. Fixed rules or lookup tables often result in unstable stacks or empty pallet space.

Trained an RL agent in Unity (C# / ML-Agents) with reward shaping for stack stability and volume density.

Exported the trained model to a C++ runtime integrated with a real industrial robot via ROS2, bridging practical differences between simulated physics and physical robot hardware.

Trained deep neural network policies in parallelized 3D Unity simulation, progressively increasing box size variety and arrival randomness.

Integrated the trained neural policy into a C++ node communicating with ROS2 topics to control physical robot arm motions.

Building a compact, reliable cocktail machine requires precise mechanical positioning without liquid spillage, hygienic fluid delivery across different viscosities, and thorough automated mixing.

Designed the full enclosure in CAD with laser-cut acrylic and 3D-printed mounts. Programmed an embedded microcontroller in C to drive a stepper-powered glass carriage along a linear track beneath peristaltic dispensers, finishing with an automated motorized DC mixer.