Every year, Google Summer of Code (GSoC) brings a fresh wave of talent into open source, and this year, three sharp, hands-on engineers are bringing their skills to Automotive Grade Linux (AGL). From AI-powered camera systems to modern Bluetooth stacks to simulated LiDAR perception, these mentees are tackling problems that sit right at the intersection of automotive engineering and open source software. Here’s a look at who they are and what they’re building.

Abdulrahman Yasser Abdulrahman — AI-Enabled Backup Camera

Abdulrahman is a full-time embedded software engineer with a Bachelor’s degree in Computer Engineering and Control Systems from Obour High Institute. His background spans C/C++, Python, OpenCV, and embedded Linux, and he’s no stranger to applying computer vision in the real world — his graduation project was a self-driving car that used YOLOv3 and CNN models running on a Raspberry Pi 4.

For GSoC 2026, Abdulrahman is bringing that computer vision expertise to AGL with an AI-Enabled Backup Camera application. The project pulls in video feeds from multiple cameras via Gstreamer pipelines and stitches them together into a seamless top-down, bird’s-eye view using OpenCV homography and warpPerspective operations. Layered on top of that is real-time object detection: a YOLOv8 or TFLite model identifies pedestrians, cars, and other obstacles as they move through the frame. The final annotated composite view is then rendered directly inside a Flutter UI on the AGL screen — turning a standard backup camera into a genuinely smart safety tool.

Seak Jian De  (Jaydon) — AGL Bluetooth Integration

Seak Jian De (Jaydon) is currently pursuing a Bachelor of Electrical-Electronics Engineering, majoring in Computer Engineering, at Universiti Teknologi Malaysia. He comes to AGL with firmware-level experience from working on SSD controllers and RISC-V development boards, providing the exact low-level systems background needed for automotive communications.

That background is put to good use in his GSoC 2026 project: AGL Bluetooth Integration. Because AGL’s legacy Bluetooth services are tied to the deprecated application framework binder (afb-daemon), Jaydon is rebuilding the stack from the ground up. The backend uses sdbus-c++ to communicate directly with BlueZ over D-Bus to manage device connections and core automotive profiles (HFP, AVRCP, PBAB, and MAP), while leveraging PipeWire and WirePlumber for dynamic audio routing. On the UI layer, the subsystem connects to AGL’s Flutter framework via bluez_native, a high-performance Dart FFI client enabling zero-copy notifications, with state managed reactively through Riverpod. The result is a Bluetooth integration built for where AGL’s architecture is headed, not where it’s been.

Shaurya Shivnath Rane — ROS-Embedded LiDAR Perception in CARLA with AGL Visualization

Shaurya is a third-year B.Tech Electronics student at VJTI in Mumbai, India, and a graduate of the Linux Kernel Mentorship Program with multiple patches already merged into the mainline Linux kernel — a strong signal of both technical depth and the persistence it takes to work in a major open source codebase.

His GSoC 2025 project, ROS-Embedded LiDAR Perception in CARLA with AGL Visualization,is bringing together simulated or real LiDAR data through ROS2 into the AGL demo UI that simulates a real vehicle ECU pipeline. A host laptop runs the CARLA simulator and publishes 3D LiDAR point cloud data over a ROS 2 bridge, while a Raspberry Pi running AGL plays the role of the vehicle ECU, receiving that point cloud data and running it through a custom Point Cloud Library (PCL) perception pipeline — voxel downsampling, RANSAC ground removal, and Euclidean clustering — to classify the obstacles around the vehicle. Those results are then visualized in real time through a custom bird’s-eye-view Flutter HMI, giving a clear, dynamic picture of what the “vehicle” perceives as it moves through the simulation.

Why This Matters

Each of these projects adds new capabilities to the AGL Unified Codebase capabilities: smarter camera systems, a Bluetooth stack that is integrated with Flutter, and A ros2 integration  and embedded hardware. Just as important, all three mentees are building on AGL’s modern Flutter-based application framework — a sign of where the platform, and the community around it, is heading.

We’re excited to support Abdulrahman, Jaydon, and Shaurya throughout their GSoC journeys, and we can’t wait to see their work land in AGL. Stay tuned for updates as these projects take shape.