Embedded Systems & Robotics Engineer | AI on the Edge | Mechatronics

Hi! I'm Arnau an engineer working at the intersection of robotics, embedded systems, and AI.
I like building real things that live on microcontrollers, sensors, and robots.

I work with:

• Embedded C/C++ · STM32 · ESP32 · Raspberry Pi · Arduino
• ROS2 · Python · AI/ML for real-time perception · Android Apps
• Control systems · RTOS · SYMULINK · MATLAB/YALMIP
• IoT: MQTT · BLE · UWB · RL systems

• Spot Arm Teleoperation
  Spot, Python, RealSense SDK, MediaPipe, Kalman Filter, Boston Dynamics SDK, Socket Networking
  Real-time teleoperation system for the Boston Dynamics Spot Arm fusing depth (RealSense) and RGB data via a Kalman Filter. Features 6DoF hand-to-end-effector mapping, gesture-based gripper control, and a decoupled client-server architecture for trajectory recording and replay.
  → spot_ag

• 3D Reconstruction
  C++, ROS2, Open3D, RealSense SDK, Point Cloud Processing
  Depth-image processing pipeline that converts RealSense data into point clouds for robotic mapping, including distance filtering and multi-frame averaging.
  → robotics_project

• Hope: Real-Time Italian Sign Language Translator Python, MediaPipe, LSTM, TensorFlow/Keras, NLP/LLM, TTS End-to-end MVP that tracks hand/pose landmarks, classifies Italian Sign Language with an LSTM model, and generates spoken Italian phrases through an LLM and TTS pipeline.
  → Hope

• Stewart Platform Kinematics & Dynamics
  MATLAB, Simulink, Kinematics, Dynamics, Control, Python
  Study of the workspace, kinematics, and dynamic response of a 6-DoF motion platform driven by four vertical linear actuators.
  → Stewart_platform

• Reinforcement Learning Experiments
  Python, Jupyter, NumPy, PyTorch/TensorFlow, RL Algorithms, Gym
  A collection of reinforcement learning exercises and projects implemented in Python and Jupyter Notebooks, covering fundamental algorithms and custom experiments.
  → reinforcement-learning

• UWB Framework (MobileKnowledge)
  C/C++, NXP QN9090, Embedded Firmware, UWB, Hardware Validation
  Firmware + validation tools for indoor positioning and robotics applications.

• IoT Voice Assistant
  Python, MQTT, Raspberry Pi, Speech Recognition, NLP, MQTT
  Voice-activated assistant on a Raspberry Pi: listens for a wake word, transcribes a command, parses intent, and publishes actions over MQTT.
  → iot-assistant

• Kinesis - Hand-Control Car
  C/C++, STM32, IMU Sensors, ChibiOS/RTOS, EKF, PWM Control, BLE
  EKF-based motion-tracking using an STM32G474 + IKS01A3 IMU to convert hand orientation into PWM motor control, with real-time RTOS scheduling and Bluetooth communication.
  → EKF-HandControl

• Slap-Chop
  C/C++, ESP32, MQTT, Grafana, Sensor Fusion, Hackathon Prototype
  Hackathon project combining ESP32 smart-gloves with MQTT streaming for gesture-based interactions. Built for HackUPC2024.
  → Slap-Chop

  Vueiting
  Python, Kotlin, MQTT, IoT Messaging, Real-Time Data Visualization
  Dual-repo setup featuring a Kotlin/Python MQTT client and a Python MQTT server for IoT messaging architecture, designed for lightweight real-time data exchange and visualization. Built for HackUPC2025. → hack25_mqtt_app → hack25_mqtt_server

• Smart knee-Sleve
  C/C++, ESP32, Python, LSTM, Sensor Fusion, UDP/Serial, GUI, Data Visualization
  Firmware for an ESP32 sensor node plus Python tools for real-time knee-angle estimation, UDP/Serial streaming and prediction ML model. Includes GUI, live plotting, and synchronized video+sensor recording.
  → smart-werables

• Designing a Spot robot controller using MediaPipe + CV for gesture-based control, extended to multi-camera environments for robust tracking and viewpoint fusion
• Improving and deepening low-level C and embedded Linux skills
• Implementing and testing new AI models to validate performance and capabilities on edge devices