Bruno Andreoni Sarmento

Developing an emotion detection feature (e.g., stress, joy) for the internal FDcore image analysis algorithm using in-car cameras. Integrating and validating this computer vision solution within an Android Automotive environment under real-world conditions. Collaborating with cross-functional teams (Embedded Software, AI, UX) in a SAFe Agile framework using C++, Java/Kotlin, and Linux to build safer, more intuitive vehicles.

Developed low-level C drivers for the ESP32 platform to control peripherals, including LEDs and a Real-Time Clock (RTC). Additionally, contributed to improving code quality by performing testing and writing comprehensive technical documentation for the firmware.

Developed a Reinforcement Learning (RL) framework for ArduPilot to automate flight parameter tuning. The system features a Lua scripting layer and C++ enhancements to the SITL simulator, enabling AI agents to successfully perform complex tasks. These included discovering energy-optimal flight paths and auto-tuning control systems by adapting to the unique dynamics of different aircraft, establishing a foundation for self-adapting algorithms in the ecosystem.

Research grant for the development of a collaborative surveillance network aimed at large urban centers. Focused on creating an intelligent system for crime prevention through the use of AI technologies, computer vision, Internet of Things, and Smart Cities, leveraging the Swarm platform for IoT.